The CCNG2 gene that encodes the unconventional cyclin G2 was one of the few genes up-regulated on anti–human epidermal growth factor receptor 2 (HER2) antibody–mediated inhibition of HER2 signaling. The purpose of this study was to explore how HER2 signaling modulates cyclin G2 expression and the effect of elevated cyclin G2 on breast cancer cell growth. Treatment of breast cancer cells that overexpress HER2 (BT474, SKBr3, and MDAMB453) with the anti-HER2 antibody trastuzumab or its precursor 4D5 markedly up-regulated cyclin G2 mRNA in vitro and in vivo, as shown by real-time PCR. Immunoblot and immunofluorescence analysis with specific antibodies against cyclin G2 showed that anti-HER2 antibody significantly increased cyclin G2 protein expression and translocated the protein to the nucleus. Trastuzumab was not able to induce cyclin G2 expression in cells weakly expressing HER2 (MCF7) or in cells that had developed resistance to trastuzumab. Enforced expression of HER2 in T47D and MDAMB435 breast cancer cells reduced cyclin G2 levels. Collectively, these data suggest that HER2-mediated signaling negatively regulates cyclin G2 expression. Inhibition of phosphoinositide 3-kinase (LY294002), c-jun NH2-terminal kinase (SP600125), and mammalian target of rapamycin (mTOR)/p70 S6 kinase (p70S6K; rapamycin) increased cyclin G2 expression. In contrast, treatment with inhibitors of p38 mitogen-activated protein kinase (SB203580), mitogen-activated protein kinase/extracellular signal–regulated kinase kinase 1/2 (U0126), or phospholipase Cγ (U73122) did not affect cyclin G2 expression. Anti-HER2 antibody in combination with LY294002, rapamycin, or SP600125 induced greater cyclin G2 expression than either agent alone. Ectopic expression of cyclin G2 inhibited cyclin-dependent kinase 2 activity, Rb phosphorylation, cell cycle progression, and cellular proliferation without affecting p27Kip1 expression. Thus, cyclin G2 expression is modulated by HER2 signaling through multiple pathways including phosphoinositide 3-kinase, c-jun NH2-terminal kinase, and mTOR signaling. The negative effects of cyclin G2 on cell cycle and cell proliferation, which occur without altering p27Kip1 levels, may contribute to the ability of trastuzumab to inhibit breast cancer cell growth. [Mol Cancer Ther 2007;6(11):2843–57]
Cyclin G2 belongs to the “G” family of unconventional cyclins correlated with cell cycle arrest that associate with active protein phosphatase 2A (1–3). G cyclins have no known active cyclin-dependent kinase (CDK) partner and, in contrast to cell cycle promoting cyclins, are up-regulated as cells undergo cell cycle arrest and apoptosis. Cyclin G family mRNA levels are relatively low in proliferating cells but significantly up-regulated during cell cycle arrest responses and highly expressed in the heart, brain, and immune system of postnatal animals (1–4). Expression of cyclins G1 and G2 is induced by a variety of genotoxic stresses and receptor-mediated growth inhibitory stimuli (2–4). Cyclins G1 and G2 have also been linked to differentiation and the development of tissues. Both are correlated with inhibition of uterine cell proliferation. Cyclin G1 is primarily associated with epithelial and stromal cell differentiation before implantation of the blastocyst, whereas cyclin G2 is associated with terminal differentiation and apoptosis of epithelial and stromal cells after implantation (5).
Cyclin G2 shares 72% homology with cyclin G1 and 41% homology with cyclin I (1, 3). In contrast to cyclin G1 and cyclin I, cyclin G2 expression fluctuates during the cell cycle with a peak level of expression in late S/early G2 phase (1–3). Cyclin G2 contains a cyclin box domain that resembles the cyclin A domain required for CDK2 interaction, implying that cyclin G2 may mediate growth inhibition by sequestering CDK2 or other CDKs (4). Yet, cyclin G2 lacks the conserved α1 and α3 helix sequences present in cyclin A required for interaction with p27Kip1 (4). Although cyclin G2 is structurally similar to cyclin A, it does not associate with cyclin-dependent kinases CDK2, CDK1, or CDK4, and cyclin G2 complexes lack typical CDK-like activity (4). The predicted inability of cyclin G2 to associate with p27Kip1 and its lack of association with CDK2 or of kinase activity against histone H1 suggest that cyclin G2 does not inhibit growth through interaction with a p27Kip1/p21WAF1–bound CDK (4, 6). Rather, like cyclin G1, cyclin G2 associates with protein phosphatase 2A (6). Whereas cyclin G1 is primarily a nuclear protein, cyclin G2 is a nucleocytoplasmic shuttling protein that primarily localizes to detergent-resistant cytoplasmic compartments and cytoskeletal elements in unperturbed cells (6, 7). Overexpression of cyclin G2 promotes the formation of aberrant lobulated nuclei, consistent with the observation that endogenous cyclin G2 associates with centrosomes and microtubules and the idea that dysregulation of its expression could promote mitotic or cytokinetic defects (6, 7). Indeed, ectopic expression of cyclin G2 inhibits proliferation of several cell types (4–8). Intriguingly, whereas DNA damage or receptor-mediated growth inhibitory stimuli induce cyclin G2 expression (2–4), activity of estrogen-bound estrogen receptor-corepressor complexes represses basal expression of cyclin G2 (9, 10). Inhibition of the phosphoinositide 3-kinase (PI3K) pathway and consequential activation of the forkhead transcription factors FoxO1 and FoxO3a directly activates CCNG2 gene expression (11, 12). Rapamycin was reported to up-regulate cyclin G2 level (13). Nevertheless, other signaling pathways that regulate cyclin G2 expression remain unclear.
The human epidermal growth factor receptor 2 (HER2; also known as ErbB-2 or c-neu) is amplified and/or overexpressed in ∼20% to 30% of breast cancer patients (14). Patients expressing high HER2 expression exhibit poor prognosis (15). The anti-HER2 antibody trastuzumab has been shown to significantly enhance the effectiveness of conventional chemotherapy for patients with breast cancers that overexpress HER2 (16). We and others have reported that anti-HER2 monoclonal antibodies exert inhibitory effects on HER2-overexpressing breast cancers through induction of p27Kip1, reduction of CDK2, and induction of G1 arrest of the cell cycle (17–23). In further elucidating the molecular mechanisms by which anti-HER2 antibodies inhibit growth of breast cancer cells that overexpress HER2, we have profiled genes that are expressed before and after treatment with trastuzumab using Affymetrix microarrays (24). We found that CCNG2 was among the few genes associated with inhibition of cell cycle progression that exhibited significantly up-regulated transcripts on trastuzumab treatment. In this report, we examine the roles of the HER2 receptor and the downstream c-jun NH2-terminal kinase (JNK), mammalian target of rapamycin (mTOR), and PI3K signaling pathways in the regulation of cyclin G2 expression and the consequences of cyclin G2 elevation to the growth of breast cancer cells. We hypothesize that multiple signaling pathways including PI3K signaling regulate cyclin G2 and that trastuzumab-induced cyclin G2 expression could contribute to the ability of this monoclonal antibody to inhibit HER2-driven proliferation of breast cancer cells.
Materials and Methods
The human breast cancer cell lines SKBr3, BT474, T47D, MDAMB453, MDAMB435, and MCF7 were obtained from the American Type Culture Collection. SKBr3, T47D, MDAMB453, and MDAMB435 cells were grown in complete medium that contained RPMI 1640 (M. D. Anderson Core Media Facility) supplemented with 10% fetal bovine serum (Sigma), 2 mmol/L l-glutamine, 100 units/mL penicillin, and 100 mg/mL streptomycin in humidified air with 5% CO2 at 37°C. BT474 and MCF7 cells were grown in complete medium containing DMEM (M. D. Anderson Core Media Facility) supplemented with 10% fetal bovine serum, 2 mmol/L l-glutamine, 1 mmol/L sodium pyruvate (Sigma), 100 units/mL penicillin, and 100 mg/mL streptomycin. For all experiments, cells were detached with 0.25% trypsin-0.02% EDTA. For cell culture, 2 × 105 to 6 × 105 exponentially growing cells were plated into 100-mm tissue culture dishes or 3 × 103 into 96-well plates in complete medium. After culture overnight in complete medium, cells were treated with anti-HER2 antibody 4D5 at 5 to 10 mg/mL or with trastuzumab at 10 mg/mL in complete medium at 37°C for 24 to 48 h. Monoclonal antibody MOPC21 served as a control for 4D5 and was used at 5 to 10 mg/mL. Human immunoglobin G served as control for trastuzumab and was used at 10 mg/mL.
Anti-HER2 murine monoclonal antibody 4D5 and humanized monoclonal antibody trastuzumab (Herceptin) were kindly provided by Genentech. MOPC21 murine myeloma cells were obtained from the American Type Culture Collection. MOPC21 cells were grown in the peritoneal cavities of BALB/c mice to produce ascites fluid and the immunoglobin was purified as previously reported (24). A control IgG1 was purchased from Calbiochem and further dialyzed against sterile cold PBS to eliminate sodium azide. Antibodies reactive with phospho-p38, total p38, phospho-Ser473/Thr308 AKT, phospho-Thr421/Ser424 p70 S6 kinase (p70S6K), total p70S6K, phospho-Thr389 p70S6K, phospho-Ser780 Rb, phospho-Thr356 Rb, and total Rb were purchased from Cell Signaling Technology, Inc. Antibodies to phospho–extracellular signal–regulated kinase (ERK) 1/2 and phospho-JNK were obtained from Promega Co. Monoclonal antibody to β-actin was obtained from Sigma. Antibodies against CDK2 and CDK4 (for Western blotting) were purchased from Santa Cruz Biotechnology, Inc. Rabbit anti–green fluorescent protein (GFP) antibodies were obtained from Abcam. Monoclonal antibody reactive with HER2 used for Western blotting was purchased from Oncogene Research Products. PI3K inhibitor LY294002, mitogen-activated protein kinase (MAPK)/ERK kinase-1/2 inhibitor U0126, and phospholipase Cγ (PLCγ) inhibitor U73122 were purchased from BIOMOL Research Laboratories, Inc. p70S6K/mTOR inhibitor rapamycin, p38 MAPK inhibitor SB203580, and JNK inhibitor SP600125 were obtained from Calbiochem-Novabiochem Corp. Transfection reagent LipofectAMINE 2000 was purchased from Life Technologies, Inc.
Cyclin G2 Antibodies
Generation of rabbit anti–cyclin G2 antibodies was described previously (6). Sheep anti–cyclin G2 antibodies were similarly prepared by immunizing animals with full-length G2-glutathione S-transferase (GST) fusion proteins; antisera tested for specificity against GST, cyclin G2-GST, and cyclin G1-GST fusion proteins; and cyclin G2–specific antibodies affinity purified essentially as described (6). Briefly, 250 μg of the affinity purified full-length cyclin G2-GST fusion protein solubilized in PBS and emulsified in an equal volume of Freund's complete (initial injection) or incomplete adjuvant were used to immunize sheep every 3 to 4 weeks. The sheep antiserum was preabsorbed with 10 mg/mL rat liver acetone powder (Sigma) for 1 h at 4°C because cyclin G2 mRNA is not detectable in the liver. Next, the antiserum was cleared by centrifugation and incubated for 1 h with GST followed by 1 h with cyclin G1-GST affinity columns to remove antibodies against the GST moiety of the fusion protein or those cross-reactive with cyclin G1. Thereafter, the antisera were incubated for 4 to 10 h at 4°C with cyclin G2-GST cross-linked with dimethyl pimelimidate to glutathione-sepharose. Finally, the affinity-purified sheep anti–cyclin G2 antibodies were eluted with 10 mL of 100 mmol/L glycine (pH 2.5) neutralized with 1 mL of 1 mol/L Tris-Cl (pH 8) and concentrated.
Generation of the Trastuzumab-Resistant BT474 Cell Line
BT474 cells at low confluency were continuously treated with trastuzumab at a concentration of 10 mg/mL for 6 months. Cultures were replenished with fresh medium containing new trastuzumab every week. When cells reached confluency, cultures were split and plated in new flasks. After 6 months, cells were checked for the ability to respond to treatment with trastuzumab with up-regulation of p27Kip1 and cell cycle arrest. Cells that were resistant to trastuzumab were then replated in cell culture dishes at very low confluency. Individual colonies were picked under the microscope, expanded, and rechecked for resistance to trastuzumab. Several subclones of BT474 cells with different levels of resistance to trastuzumab have been identified. The subclone used in this study exhibited maximal resistance to the antibody.
Generation of T47D and MDAMB435 Stable Cell Lines That Overexpress HER2
T47D or MDAMB453 cells were transfected with a vector containing full-length human HER2 cDNA and stable clones expressing high levels of HER2 were selected as described in a previous report (24).
Establishment of Inducible Cyclin G2-GFP Stable Clones in MCF7 Cells
Ponasterone/muristerone inducible expression of GFP-tagged cyclin G2 was established in MCF7 cells using the ecdysone-inducible expression system (25) and the receptor expression plasmid pVgRXR, inducible receptor target control and expression vectors pIND/LacZ and pIND, and protocols supplied by the manufacturer (Invitrogen). MCF7 cells expressing the ecdysone and retinoid X receptors (RXR) were first established by introduction of MluI-linearized pVgRXR DNA into the cells followed by selection for growth in the presence of the antibiotic Zeocin. Zeocin-resistant clones were tested for the expression of ponasterone-responsive receptors by transient transfection with the pIND/LacZ control plasmid followed by detection of ponasterone-inducible, promoter-driven LacZ expression. Cyclin G2-GFP (6) was subcloned into pIND, and those clones with the best and tightest inducibility were then expanded, transfected, and selected for the presence of integrated linearized pIND/G2-GFP by selection with G418. G418-resistant clones were screened for ponasterone-inducible (1–10 μmol/L) expression of tagged cyclin G2 by fluorescence microscopy analysis of GFP expression. Two clones were further tested for inducible (10 μmol/L ponasterone) cyclin G2 and GFP expression by immunoblotting and flow cytometry analysis as previously described (6, 7).
Preparation of Total RNA
Total RNA was extracted by using TRIzol reagent (Invitrogen). Procedures were done according to the manufacturers' recommendation. The purity of RNA was assessed by absorption at 260 and 280 nm (values of the ratio of A260/A280 of 1.9–2.1 were considered acceptable) and by ethidium bromide staining of 18S and 28S RNA on gel electrophoresis. RNA concentrations were determined at A260.
Quantitative Real-time Reverse Transcription-PCR Analysis
To detect cyclin G2 expression, quantitative real-time reverse transcription-PCR (RT-PCR) analysis was done with an ABI Prism 7900HT Sequence Detection System using TaqMan universal PCR master mix according to the manufacturer's specifications (ABI). The TaqMan probe and primer used for cyclin G2 was assay ID no. Hs00171119_m1. The human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene was used as an endogenous control (ABI). The gene-specific probes were labeled with reporter dye FAM and the GAPDH internal control probe was labeled with a different reporter dye (VIC) at the 5′ end. A nonfluorescent quencher and the minor groove binder were linked at the 3′ end of probe as quenchers. The thermal cycler conditions were as follows: hold for 10 min at 95°C, followed by two-step PCR for 40 cycles of 95°C for 15 s followed by 60°C for 1 min. All measurements were done in triplicate. Amplification data were analyzed with an ABI Prism Sequence Detection Software version 2.1 (ABI). To normalize the relative expression of the gene of interest to the GAPDH control, standard curves were prepared for each gene studied and for the GAPDH in each experiment. When the efficiency of the target gene amplification and the efficiency of GAPDH amplification were approximately equal, which was proved by examining the absolute value (<0.1) of the slope of log input amount versus δCT, the δδCT method recommended by the manufacturer was used to compare the relative expression levels between treatments. When the efficiency of the target gene amplification and the efficiency of GAPDH amplification were not equal, a Pfaffl's relative expression model was used to calculate the relative expression levels of samples as reported previously (24).
BT474 Xenografts in Nude Mice
BT474 xenografts were established in female nu/nu BALB/c mice as described previously (26). Briefly, mice with established BT474 tumors were treated with trastuzumab at 1 mg/kg or control human IgG at 1 mg/kg twice per week for 3 weeks. Total RNA was extracted as described above for cyclin G2 detection.
Detection of Cyclin G2 Protein by Immunoprecipitation and Immunoblot Analysis
These procedures were done as described previously (6, 7). SKBR3 and MCF-7 cells were lysed in radioimmunoprecipitation assay buffer (10% glycerol, 1% NP40, 150 mmol/L NaCl, 5 mmol/L EDTA, 5 mmol/L EGTA, 50 mmol/L Tris, pH 7.4) containing 0.4% deoxycholate, 0.05% SDS, and protease inhibitors [pepstatin A (1 μg/mL), leupeptin (10 μg/mL), aprotinin (20 μg/mL), and phenylmethanesulfonly fluoride (200 nmol/L)]. Cell lysates were centrifuged at 10,000 × g to remove insoluble material. A 1-mg aliquot of cleared lysate was sequentially incubated for 1 h each with 10, 5, and 0 μg of sheep IgG plus protein A- and protein G-sepharose, followed by 4-h incubation with 10 μg of sheep anti–cyclin G2 antibody in 20 μL of protein G and protein A-sepharose beads (1:2 ratio). Cyclin G2 and control IgG immunoprecipitates were washed twice in radioimmunoprecipitation assay buffer containing 500 mmol/L NaCl, twice in radioimmunoprecipitation assay buffer alone, and once in double-distilled water. Immunoprecipitated proteins were extracted with SDS sample buffer and run on a SDS-PAGE gel. Western blot analysis was done with affinity-purified polyclonal rabbit anti–cyclin G2 antibodies as previously described (6, 7).
Cyclin G2 Immunofluorescence Staining
SKBr3 cells (1.5 × 105) were seeded onto collagen (10 μg/mL)– and poly-l-lysine (1 μg/mL)–coated 22-mm2 glass coverslips in six-well dishes. Cells were grown for 14 to 18 h and then either cultured untreated or treated with control MOPC21 (5 μg/mL), 4D5 (5 μg/mL), or trastuzumab (10 μg/mL) antibodies for 24 h. Cells were fixed in ice-cold methanol at −20°C for 6 to 8 min. Fixed cells were permeabilized and immunostained with polyclonal rabbit anti–cyclin G2 antibodies (6) and DM1A mouse anti–α-tubulin monoclonal antibodies from NeoMarkers, followed by incubation with FITC-conjugated donkey anti-rabbit and rhodamine-conjugated donkey anti-mouse secondary antibodies (both from Jackson ImmunoResearch Laboratories). Coverslips were mounted on glass slides and confocal microscopy images were collected with a Bio-Rad MRC 1024 microscope.
Flow Cytometric Analysis and Sorting
These procedures were done as described previously (7, 24).
Anchorage-Dependent Cell Growth Assay
A crystal violet cell growth assay in 96-well microtiter plates was used to assess anchorage-dependent growth as described previously (26).
All experiments were repeated at least thrice on different occasions. The results are presented as the mean ± 95% confidence intervals for all values. A paired Student's t test was used to evaluate statistically significant differences in cyclin G2 levels between the treatment groups and the vehicle control group. P < 0.05 was considered statistically significant. All statistical tests and corresponding P values were two sided.
Cyclin G2 Expression Is Increased in Breast Cancer Cells That Overexpress HER2 by Treatment with Anti-HER2 Antibody
CCNG2 was initially identified by Affymetrix microarray analysis as one of the genes that can be up-regulated by trastuzumab treatment of breast cancer cells that overexpress HER2 (24). Quantitative real-time RT-PCR for cyclin G2 mRNA was then done in the SKBr3, BT474, and MDAMB453 cell lines that overexpress HER2 to validate results obtained from microassay studies. As shown in Fig. 1A, treatment with anti-HER2 antibody 4D5, the precursor of the humanized antibody trastuzumab, induced cyclin G2 expression in a dosage-dependent manner. Trastuzumab was capable of rapidly increasing cyclin G2 expression as early as 4 h after treatment in BT474 cells (Fig. 1A). A time-dependent increase in cyclin G2 expression was confirmed in MDAMB453 cells (Fig. 1A). To determine whether induction of cyclin G2 protein expression paralleled the up-regulation of cyclin G2 mRNA, we carried out immunoblot analysis of cyclin G2 immunocomplexes with two different affinity-purified anti–cyclin G2 antibodies (4, 6). As shown in Fig. 1B, cyclin G2 protein was clearly up-regulated in cells treated with trastuzumab (10 μg/mL) for 24 h, but not in untreated control SKBr3 cells. Next, we carried out dose- and time-response experiments to assess the effect of anti-HER2 antibody on up-regulation of cyclin G2 protein. Cyclin G2 levels increased with increasing concentrations of 4D5 (1–10 μg/mL) during treatment of SKBr3 cells for as short as 8 h (Fig. 1C). In contrast, control MOPC21–treated cells did not show any increased expression of cyclin G2 (Fig. 1C). Similar to trastuzumab, 4D5 treatment for 24 h also up-regulated protein expression of cyclin G2 (Fig. 1C). Our previous studies showed that trastuzumab inhibited tumor growth in vitro and in vivo (26, 27). Cyclin G2 levels in BT474 xenografts after trastuzumab treatment were also assessed. As shown in Fig. 1D, trastuzumab clearly induced cyclin G2 RNA expression in vivo, consistent with the in vitro results. Taken together, these data suggest that targeting the HER2 pathway with anti-HER2 monoclonal antibody increases cyclin G2 expression at both the RNA and protein levels.
The HER2 Pathway Regulates Cyclin G2 Expression
The effects observed with anti-HER2 antibodies suggest that cyclin G2 levels might be regulated through HER2 signaling. To further confirm this relationship, breast cancer cell lines with low levels of HER2 expression were transfected with expression vectors that contained or lacked HER2. Levels of cyclin G2 were determined by quantitative real-time RT-PCR in the pairs of breast cancer cell lines with low and high HER2 expression. As shown in Fig. 2A, T47D or MDAMB435 cells that overexpress HER2 had lower expression of cyclin G2 when compared with cells transfected only with an empty vector. MCF7 cells endogenously express low levels of HER2 and are therefore not sensitive to growth inhibition after trastuzumab treatment (Fig. 2B). As shown in Fig. 2C, trastuzumab did not induce cyclin G2 expression in MCF7 cells. Similarly, as shown in Fig. 2D, trastuzumab did not increase cyclin G2 expression in a BT474 subline that had developed resistance to trastuzumab (Fig. 2B). Thus, cyclin G2 expression correlates inversely with active HER2 signaling.
An Additive Increase in Cyclin G2 Expression Is Produced by Treatment with Anti-HER2 Antibody and an Inhibitor of the PI3K Pathway in Breast Cancer Cells That Overexpress HER2
Our previous studies have shown that trastuzumab inhibits growth of cancer cells that overexpress HER2 by affecting several signaling pathways, including PI3K, p70S6K, p38 MAPK, and JNK (24, 26, 27). The CCNG2 gene contains two forkhead response elements recognized by the FoxO transcription factors and has been reported to be negatively regulated downstream of the PI3K-AKT-FoxO signaling pathway (11, 12). To evaluate the role of different signaling pathways in regulating expression of cyclin G2, we initially confirmed the importance of PI3K. As expected, the PI3K inhibitor LY294002 markedly increased cyclin G2 expression in BT474 cells (Fig. 3A) and SKBr3 cells (Fig. 3B). Interestingly, when these cells were treated with a combination of anti-HER2 antibody (trastuzumab or 4D5) and the PI3K inhibitor LY294002, greater expression of cyclin G2 was observed than with either single agent (Fig. 3A and B). In contrast, control human IgG in combination with LY294002 did not induce additional expression of cyclin G2 in either BT474 or SKBr3 cells (Fig. 3A and B). In these studies, LY294002 inhibited phosphorylation of AKT at Ser473 and Thr308 on Western blot analysis (Fig. 3C). LY294002 inhibited cell proliferation without inducing apoptosis in SKBr3 cells (Fig. 3D). Our data confirm that the PI3K pathway regulates levels of cyclin G2 expression and suggest that not all of the effects of anti-HER2 antibody are mediated through PI3K signaling.
An Additive Increase in Cyclin G2 Expression Is Produced by Treatment with Anti-HER2 Antibody and an Inhibitor of the mTOR Pathway in Breast Cancer Cells That Overexpress HER2
As shown in Fig. 4A, application of rapamycin, an inhibitor of mTOR and consequently of p70S6K, induced a more robust expression of cyclin G2 in BT474 cells than did trastuzumab. Similarly, expression of cyclin G2 was markedly up-regulated by rapamycin in SKBr3 cells (Fig. 4B). Importantly, combinatorial treatments of anti-HER2 antibody (trastuzumab or 4D5) plus rapamycin induced greater expression of cyclin G2 than did either agent alone in both BT474 and SKBr3 cells (Fig. 4A and B). In contrast, control human IgG in combination with rapamycin did not induce additional expression of cyclin G2 in either BT474 or SKBr3 cells (Fig. 4A and B). The ability of rapamycin to inhibit the phosphorylation of p70S6K at Thr389 was verified by Western blot analysis (Fig. 4C). Rapamycin inhibited cell proliferation without inducing apoptosis in SKBr3 cells (Fig. 4D). Thus, the mTOR/p70S6K pathway seems to be involved in the regulation of cyclin G2 expression. Moreover, levels of cyclin G2 can be further induced with a combination of anti-HER2 antibody and inhibitors of mTOR.
An Additive Increase in Cyclin G2 Expression Is Produced by Treatment with Anti-HER2 Antibody and an Inhibitor of the JNK Pathway in Breast Cancer Cells That Overexpress HER2
A JNK inhibitor SP600125 was also found to affect cyclin G2 expression. As shown in Fig. 5A, treatment with SP600125 increased cyclin G2 expression in BT474 cells to levels similar to those observed after treatment with trastuzumab. Combined treatment with trastuzumab and SP600125 produced greater expression of cyclin G2 than did either single agent (Fig. 5A). In SKBr3 cells, SP600125 alone induced less cyclin G2 expression (Fig. 5B), but anti-HER2 antibody 4D5 plus SP600125 induced much higher levels of cyclin G2 expression than either single agent (Fig. 5B). In contrast, control human IgG in combination with SP600125 did not induce additional expression of cyclin G2 in either BT474 or SKBr3 cells (Fig. 5A and B). The ability of SP600125 to inhibit JNK phosphorylation was verified by Western blot analysis (Fig. 5C). SP600125 inhibited cell proliferation by inducing apoptosis in SKBr3 cells (Fig. 4D). Thus, cyclin G2 can be regulated by the JNK pathway, albeit to a lesser degree than by the PI3K or p70S6K pathways. Moreover, increased expression of cyclin G2 was observed with a combination of JNK and anti-HER2 antibody.
Cyclin G2 Expression Is Also Regulated by PI3K, JNK, and p70S6K in Breast Cancer Cells with Normal Levels of HER2
To test the roles of the PI3K, JNK, and p70S6K pathways in regulating cyclin G2 expression in human breast cancer cell lines that express low levels of HER2, MCF7 and T47D cells were treated with the inhibitors of PI3K, JNK, and p70S6K. As shown in Fig. 6A, LY294002, SP600125, and rapamycin all induced RNA expression of cyclin G2 in MCF7 cells in a dosage-dependent manner. Similar results were seen in T47D cells (data not shown). Consistent with the data presented in Figs. 3–5, LY294002 produced the highest levels of cyclin G2, and SP600125 is the weakest inducer of cyclin G2 (Fig. 6A). Cyclin G2 protein levels in SKBr3 cells (high HER2) and MCF7 cells (low HER2) after LY294002 treatment were measured by immunoprecipitation-Western blotting. As shown in Fig. 6B, LY294002 induced marked cyclin G2 expression at protein level in both SKBr3 and MCF7 cells. The baseline protein levels in MCF7 cells that express low HER2 were higher than those in SKBr3 cells that overexpress HER2, which is consistent with the RNA levels of cyclin G2 shown in Fig. 2A. Taken together, these data further confirm the roles of HER2, PI3K, JNK, and p70S6K pathways in regulating cyclin G2 expression in human breast cancer cells.
Cyclin G2 Expression Is Not Affected by Inhibition of the ERK-MAPK, p38 MAPK, or PLCγ Pathway
Whereas our data show regulation of cyclin G2 by the PI3K, JNK, and mTOR/p70S6K pathways in human breast cancer cell lines, other pathways do not seem to be integrally involved. Inhibitors of p38 MAPK (SB203580), MAPK/ERK kinase 1/2 (U0126), and PLCγ (U73122) did not affect cyclin G2 expression significantly. As shown in Fig. 6C and D, SB203580 did not dramatically increase cyclin G2 expression in both BT474 and SKBr3 cells. Unlike the inhibitors of PI3K, JNK, and p70S6K pathways, combined treatment with SB203580 and anti-HER2 antibody trastuzumab or 4D5 failed to significantly induce cyclin G2 expression (Fig. 6C and D). Similar results were seen with U0126 and U73122 (data not shown). These data suggest that ERK MAPK, p38 MAPK, and PLCγ pathways do not regulate cyclin G2 expression. The ability of SB203580 to inhibit p38 phosphorylation was verified by Western blot analysis (Supplementary Fig. S1A).3
Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
Ectopic Expression of Cyclin G2 Inhibits CDK2 Activity and Rb Phosphorylation, Hindering Cell Cycle Progression and Cellular Proliferation
To study the effect of elevated cyclin G2 in trastuzumab-treated breast cancer cells that overexpress HER2, expression plasmids encoding GFP-tagged cyclin G2 were transfected into SKBr3 cells. Cyclin G2–transfected SKBr3 cells were then analyzed and fluorescence-activated cell sorted into GFP-positive and GFP-negative fractions for immunoblot analysis. As shown in Fig. 7A, ectopic expression of cyclin G2 did not alter p27Kip1 and CDK4 expression. However, overexpression of cyclin G2 markedly reduced the active form of CDK2 (Fig. 7A). Overexpression of cyclin G2 depressed the phosphorylation of Rb protein at two important sites, Thr356 and Ser780 (Fig. 7A). To study the effect of cyclin G2 expression on SKBr3 cell proliferation, cells from the GFP-positive and GFP-negative fractions were replated in cell culture dishes and continuously cultured for 7 days. As shown in Fig. 7B, the cells containing cyclin G2 (GFP+) grew slower than the control cells that had not been transfected or the cells that contained an empty vector.
Attempts to establish the stable clones that inducibly express ectopic cyclin G2 in SKBr3 cells were not successful. However, stable clones that express ponasterone-inducible cyclin G2 under the control of the hybrid ecdysone response element–driven promoter in MCF7 cells were established as described in Materials and Methods. As shown in Fig. 7C, clones 26 and 27 inducibly expressed different, but relatively modest, levels of exogenous cyclin G2 protein when cultured in the presence of 10 μmol/L ponasterone for 48 h. Both levels of ponasterone-inducible cyclin G2-GFP expression achieved through the hybrid ecdysone response element–driven promoter in stable MCF7 clones and of cyclin G2-GFP expression produced from a constitutive cytomegalovirus promoter–driven promoter in MCF7 cells transiently transfected with pcDNA3-cyclin G2-GFP were assessed by flow cytometry. The relative intensity of ponasterone-inducible expression of GFP-tagged cyclin G2 in clones 26 and 27 (Supplementary Fig. S2A and B)3 was determined to be >10-fold lower than that achieved by transient transfection with pcDNA3-cyclin G2-GFP (Supplementary Fig. S2C).3 Even at these low levels of cyclin G2 expression, ponasterone-induced clone 26 exhibited a significant G1-phase cell cycle arrest (Fig. 7D, top). At slightly higher levels of cyclin G2 induction, clone 27 exhibited even more significant G1-phase arrest (Fig. 7D, bottom). In contrast, uninduced control MCF-7 cells were not hindered in cell cycle progression (Fig. 7C and D). These data suggest that elevation of cyclin G2 expression in human breast cancer cells inhibits active CDK2 and Rb phosphorylation, resulting in inhibition of cell cycle progression and cellular proliferation.
Overexpression of Cyclin G2 Produces Aberrant Nuclei
Prior studies have shown that overexpression of cyclin G2 in HEK293 and Chinese hamster ovary cells induces lobulated aberrant nuclei (6). Such changes in nuclear morphology did not directly reflect apoptotic nuclei and were largely terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) negative with intact lamin B–positive nuclear envelopes (6). In this study, lobulated nuclei were detected by 4′,6-diamidino-2-phenylindole staining in cyclin G2–expressing SKBr3 cells that were marked by GFP (Fig. 8A). In contrast, there was no obvious change in nuclear morphology in control cells transfected with empty vector (Fig. 8A) or GFP-negative cells (data not shown).
Anti-HER2 Antibody Increases Nuclear Localization of Cyclin G2
Data in Figs. 1–6 have shown that treatment of breast cancer cells with the anti-HER2 antibodies trastuzumab and 4D5 up-regulates cyclin G2 expression. Recently, cyclin G2 protein has been shown to be a centrosome-associated and nucleocytoplasmic shuttling protein (7). To visualize the cellular localization of increased cyclin G2 induced by HER2 inhibitory antibodies, SKBr3 cells were treated with 4D5 or the MOPC21 control antibody for 24 h. Cells were then costained with an anti–cyclin G2 antibody labeled with an Alexa 488 (green fluorescence)–conjugated secondary antibody and an anti–α-tubulin antibody labeled with an Alexa 568 (red fluorescence) secondary antibody. As shown in Fig. 8B, endogenous cyclin G2 signals were weak and in a perinuclear pattern in control MOPC21–treated SKBr3 cells. In contrast, 4D5 treatment resulted in elevated cyclin G2 signals accumulating in somewhat speckled pattern within nuclei of treated cells (Fig. 8B). To further confirm these observations, SKBr3 cells were treated with trastuzumab for 24 h. In untreated SKBr3 cells, endogenous cyclin `G2 was localized to small dot-like structures at microtubule organizing centers indicative of centrosomes (Fig. 8C, yellow arrows). After treatment with trastuzumab, intensified cyclin G2 signals were obvious within nuclei (Fig. 8C) and similar to the distinct pattern observed in 4D5-treated cells (Fig. 8B). Increased cyclin G2 signals were also observed at some centrosomes after trastuzumab treatment (Fig. 8C). Together, these data revealed that treatment of HER2-positive breast cancer cells with anti-HER2 inhibitory antibodies significantly increases expression and nuclear localization of cyclin G2.
In this study, we have reported for the first time that treatment of breast cancer cells that overexpress HER2 with the anti-HER2 antibody induces expression of cyclin G2, resulting in translocation of accumulated cyclin G2 to the nucleus. Inhibition of several signaling including HER2, JNK, PI3K, and mTOR/p70S6K (Figs. 1–5) pathways up-regulates cyclin G2 expression. Moreover, inhibition of JNK, PI3K, or mTOR/p70S6K potentiates trastuzumab-induced cyclin G2 expression. Our data also indicate that inhibition of p38 MAPK, MAPK/ERK kinase 1/2, and PLCγ signaling does not modulate cyclin G2 expression. Finally, we showed that ectopic expression of cyclin G2 reduces expression of active CDK2, inhibits Rb phosphorylation and cell proliferation, and induces G1-phase arrest cell cycle (Fig. 7).
Consistent with a previous report, we have found that the PI3K inhibitor LY294002 induces cyclin G2 expression (Fig. 3). Overexpression of the p110a catalytic subunit of PI3K decreased expression of cyclin G2 compared with the control (12). The PI3K-AKT pathway negatively regulates the FoxO subfamily of forkhead transcription factors, which include FoxO1a, FoxO3a, and FoxO4 (29, 30). PI3K-AKT activation promotes FoxO phosphorylation and export from the nucleus into the cytoplasm (31). FoxO3a and FoxO1 activate transcription of cyclin G2 gene and increase its mRNA expression through a FoxO binding site in cyclin G2 promoter area (12). In agreement with a previous report (12), our work indicates that the ERK/MAPK pathway does not affect cyclin G2 expression.
Induction of p27Kip1 was found to be one of the critical mechanisms through which trastuzumab induces a G1-phase cell cycle arrest (21, 22, 27). Modulation of the PI3K, mTOR/p70S6K, and JNK signaling arms downstream of HER2 seems to be particularly important for anti-HER2 antibody activity (17, 23, 24). Trastuzumab-mediated down-regulation of CDK2 activity has been attributed to modulation of p27Kip1, PI3K, and mTOR activities (17, 27, 32). In this study, we found that application of anti-HER2 antibodies to cultured HER2-positive breast cancer cell lines could also up-regulate cyclin G2 expression, which, when ectopically expressed, is sufficient to inhibit CDK2, Rb phosphorylation, cell cycle progression, and cellular proliferation (Fig. 7). Importantly, ectopic cyclin G2 expression does not elevate p27Kip1 expression (Fig. 7), suggesting that the cyclin G2–mediated growth inhibitory effect does not require p27Kip1 up-regulation (1, 4, 6). Therefore, enhancement of cyclin G2 expression may be a parallel mechanism contributing to trastuzumab-mediated inhibition of cell cycle progression and cellular proliferation.
To further define the role of cyclin G2 induction in the action of trastuzumab, we carried out small interfering RNA–mediated cyclin G2 knockdown experiments. Results showed that cyclin G2 small interfering RNA could reduce the level of basal cyclin G2 mRNA by ∼70% (Supplementary Fig. S3A).3 Consistent with our earlier results, 4D5 increased cyclin G2 level. The effect of 4D5 on cyclin G2 expression was resilient even in the presence of cyclin G2 small interfering RNA and could induce some elevation of cyclin G2 (Supplementary Fig. S3A).3 Suppression of cyclin G2 mRNA expression only modestly decreased anti-HER2 antibody–mediated growth inhibition (Supplementary Fig. S3B).3 Although these results suggest that cyclin G2 up-regulation is not required for anti-HER2 antibody–mediated growth inhibition, they relied on measurement of cyclin G2 mRNA levels and did not assess cyclin G2 protein levels. It is possible that cyclin G2 is a relatively stable protein, particularly in conditions of anti-HER2 antibody treatment when it seems to be relocalized to subnuclear compartments; thus further, more detailed protein expression studies are required. As we have previously shown, p27Kip1 is the major target up-regulated by anti-HER2 antibody–mediated inhibition of HER2 (27). Cyclin G2 may only play a modulatory role in the mechanisms of actions of anti-HER2 antibody. In this respect, cyclin G2 might only contribute to trastuzumab growth-inhibitory effects through the p27Kip1-independent mechanism. It will be interesting to determine whether knockdown of cyclin G2 expression in concert with p27Kip1 depletion has a greater progrowth effect on trastuzumab-treated cells than loss of either protein alone. Clearly, more extensive, in-depth studies are necessary to elucidate the contributions of elevated cyclin G2 expression to modulation of cell cycle progression during growth-inhibitory responses.
Trastuzumab has been shown to regulate the PI3K, JNK, and mTOR/p70S6K pathways (17, 21, 22, 24, 32), all of which seem to contribute to increased cyclin G2 expression. Trastuzumab acts on the PI3K, JNK, and p70S6K pathways by directly blocking HER2 signaling. The additive effects observed with trastuzumab and specific inhibitors of PI3K, JNK, and mTOR (Figs. 3–5) suggest that HER2 signaling is only one of several factors that regulate these pathways. From a clinical perspective, more effective inhibition of cancer growth might be achieved by combining trastuzumab with specific inhibitors of each pathway. Currently, newer inhibitors of PI3K (PX-866), AKT (QLT0394 and PCK412), and p70S6K/mTOR (CCI-779, RAD-001, and AP-23573) are being evaluated in clinical trials (33). Inhibitors of the JNK pathway (CEP-1347 and SP600125) have been tested in vivo and are also entering clinical trials (34). Thus, it is worthwhile to test the effects of combination of trastuzumab with one or more of these inhibitors in future clinical trials.
Ectopic expression of cyclin G2 induces nuclear aberrations along with cell cycle arrest (6). The exact mechanism by which ectopic cyclin G2 induces the formation of aberrant, multinucleated cells is not clear, but it may relate to potential promotion of mitotic or cytokinetic defects (6), a notion consistent with the observation that endogenous cyclin G2 and protein phosphatase 2A associate with centrosomes and microtubules and that ectopic G2 expression induces microtubule bundling and resistance to depolymerization (7). Disruption of the microtubule network has been shown to generate distortions in the nuclear envelope and could cause changes in nuclear shape (35, 36). Our current study confirmed these earlier observations, extending the ability of even modestly elevated cyclin G2 to inhibit cell cycle progression (Fig. 7) in breast epithelial cells. Importantly, here we determined that inhibition of HER2 signaling induces accumulation of elevated cyclin G2 in the nucleus with a distinct subnuclear distribution (Fig. 8).
Although cyclin G2 is associated with centrosomes and microtubules in the cytosol, cyclin G2 has been shown to be a nucleocytoplasmic shuttling protein that can translocate to the nucleus (7). This is clearly shown in Fig. 8 where elevated G2 is most prominent in the nucleus of anti-HER2 antibody–treated cells. The mechanisms by which anti-HER2 antibody induces this significant increase in cyclin G2 nuclear localization are not clear, but could include a modification or new association of cyclin G2 that precludes Crm1-mediated nuclear export of cyclin G2 (7). Cyclin G2 translocation from the cytosol to the nucleus may facilitate the activity of cyclin G2 complexes on unknown nuclear targets that regulate cellular proliferation. Cyclin G1, the closest homologue of cyclin G2, was reported to localize to replication foci where it is hypothesized to associate with proliferating cell nuclear antigen (38). Interestingly, anti-HER2 therapeutics has been shown to affect genes of replication and DNA repair, including proliferating cell nuclear antigen (24). Therefore, it is possible that nuclear cyclin G2 as a homologue of cyclin G1 may localize at replication foci where it may serve as one of the effectors of anti-HER2 antibody. However, the major staining pattern of nuclear cyclin G2 seems to be particularly similar to the distribution of two subnuclear compartments of promyelocytic leukemia nuclear body (39) and the nucleolus (40). Promyelocytic leukemia nuclear body is related to the death domain–associated protein (Daxx), which has been reported to interact with a number of proteins involved in cell survival, differentiation, and apoptosis (41). Cyclin G2 has been shown to interact with nuclear proteins involved in the growth inhibitory pathway including p53, protein phosphatase 2A, and murine double minute-2 (6, 7, 37) and is induced by apoptotic stimuli in lymphocytes (4). Thus, cyclin G2 may associate Daxx or other promyelocytic leukemia nuclear body proteins during apoptosis. Nucleoli are among the most prominent subnuclear structures and are known to be important for ribosome biogenesis, gene transcription, and apoptosis (40, 42). Cyclin G1 can associate with the nucleolar proteins such as the alternative reading frame proteins (ARF) such as p14/p19 and relocalize to the nucleus when coexpressed with ARF (37). Cyclin G2 may also relocalize to the nucleolus. Indeed, our recent immunofluorescence microscopy studies of LY294002-treated MCF-7 and SKBr3 cells indicate a colocalization of up-regulated endogenous cyclin G2 with the nucleolar markers fibrillarin and nucleolin.4
A.S. Arachchige-Don and M.C. Horne, unpublished data.
Grant support: National Cancer Institute grant CA39930 (R.C. Bast) and U.S. Army Medical Research and Materiel Command grant BC045656 (M.C. Horne).
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.
Note: M.F. Horne and R.C. Bast shared equally in senior authorship.
We thank Wendy Schober at M. D. Anderson Cancer Center for her expert assistance with flow cytometric cell sorting. This project used the Flow Cytometry Core Laboratory, Media Preparation Core Facility and Animal Core Facility, supported, in part, by a Cancer Center Support Grant from the National Cancer Institute, Department of Health and Human Services to M. D. Anderson Cancer Center. We also thank Genentech (South San Francisco, CA) for providing trastuzumab and 4D5 antibodies; Jennifer Duven of the University of Iowa for establishing cyclin G2-inducible MCF7 clones; Justin Fishbaugh and Gene Hess of the University of Iowa Holden Comprehensive Cancer Center Flow Cytometry Facility for assistance with flow cytometry; and the University of Iowa Central Microscopy Facility for use of confocal microscope.