|Home | About | Journals | Submit | Contact Us | Français|
Ansamycin antibiotics, such as 17-allylaminogeldanamycin (17-AAG), bind to Hsp90 and regulate its function, resulting in the proteasomal degradation of a subset of signaling proteins that require Hsp90 for conformational maturation. HER2 is a very sensitive target of these drugs. Ansamycins cause RB-dependent G1 arrest that is associated with loss of D-cyclins via a PI3 kinase, Akt dependent pathway. Downregulation of D-cyclin was due, in part, to loss of Akt expression in response to drug. Moreover, in HER2 overexpressing breast cancer cells, 17-AAG caused rapid inhibition of Akt activity prior to any change in Akt protein. Ansamycins caused rapid degradation of HER2 and a concomitant loss in HER3 associated PI3 kinase activity. This led to a loss of Akt activity, dephosphorylation of Akt substrates, and loss of D-cyclin expression. Introduction into cells of a constitutively membrane bound form of PI3 kinase prevented the effects of the drug on Akt activity and D-cyclins. Thus, in breast cancer cells with high HER2, Akt activation by HER2/HER3 heterodimers is required for D-cyclin expression. In murine xenograft models, non-toxic doses of 17-AAG markedly reduced the expression of HER2 and phosphorylation of Akt and inhibited tumor growth. Thus, pharmacological inhibition of Akt activation is achievable with ansamycins and may be useful for the treatment of HER2 driven tumors.
Heat Shock Protein 90 (Hsp90) is an abundant chaperone that plays a role in protein refolding and in the conformational maturation of certain signaling proteins. It contains a conserved pocket in its amino-terminus that binds ATP and ADP with low affinity (Prodromou et al., 1997). Completion of Hsp90-dependent protein refolding is ATP-dependent and involves dissociation of the renaturated protein from the chaperone complex. Ansamycin antibiotics and radicicol are natural products that bind to the Hsp90 pocket and alter its function (Schneider et al., 1996). Exposure of cells to these compounds results in the degradation of a subset of signaling proteins that associate with Hsp90, including steroid receptors and the Raf serine kinase, as well as certain transmembrane tyrosine kinases, including HER2 and met (Czar et al., 1997; Mimnaugh et al., 1996; Schulte et al., 1996; Stancato et al., 1997; Webb et al., 2000; Whitesell and Cook, 1996; Xu et al., 2000). Breast cancer cells that overexpress HER2 are especially sensitive to the anti-proliferative activity of these drugs (P Munster, unpublished data). Based on these and other data, an ansamycin derivative, 17-allylaminogeldanamycin (17-AAG), is being tested in patients with advanced cancer.
The response of tumor cells to ansamycins is dependent upon the status of their RB protein. Ansamycins cause cells with wild-type retinoblastoma protein (RB) to undergo growth arrest in G1 and, subsequently, to undergo apoptosis. G1 arrest is associated with rapid loss in D-cyclin associated kinase activity and hypophosphorylation of RB. Tumor cells with defective RB-function are resistant to induction of G1 block by ansamycins (Srethapakdi et al., 2000). These data imply that effects of ansamycins on G1 progression are confined to pathways upstream of RB and that Hsp90 selectively regulates these pathways.
The sole known substrate of D-cyclin associated kinase is RB, so it is likely the downregulation of the activity of this enzyme by ansamycins explains their effect on G1 progression. These drugs are not direct inhibitors of D-cyclin-cdk4/6 protein kinase but they cause a rapid loss in its intracellular activity associated with decreased expression of D-cyclins and of cdk4. The maturation of cdk4 is dependent upon Hsp90 and this kinase is a direct target of the drug (Stepanova et al., 1996). In contrast, ansamycins decrease D-cyclin levels by downregulating a PI3 kinase, Akt dependent pathway required for their expression (Muise-Helmericks et al., 1998).
The importance of the phosphatidylinositol 3-kinase (PI3 kinase), Akt kinase pathway in maintaining the growth and viability of cancer cells led us to investigate the mechanism through which ansamycins downregulate its function. We found that 17-allylaminogeldanamycin (17-AAG) inhibited Akt function in a complex manner. It caused a slow decline in the expression of Akt protein. This decline was accompanied by a parallel diminution in Akt phosphorylation and kinase activity. However, in breast tumor cells with high levels of HER2, 17-AAG also caused a rapid loss in Akt kinase activity prior to any change in protein expression. This was accompanied by a decrease in the phosphorylation of Akt substrates and a decline in cyclin D1 and D3 expression. The loss of Akt kinase activity was due to rapid dephosphorylation of HER3 in response to degradation of HER2. This led to an uncoupling of the binding of active PI3 kinase to HER3 at the membrane. Introduction into cells of a membrane bound, prenylated form of the catalytic subunit of PI3 kinase rendered the cells insensitive to the effects of 17-AAG on Akt kinase activity and D-cyclin expression. These results suggest that in breast cancers with high levels of HER2, Akt activation is dependent on HER2/HER3 heterodimers and is required for D-cyclin expression. 17-AAG inhibits this pathway by inducing HER2 degradation and by reducing Akt expression, so it may be useful in treating these tumors. Indeed, we showed here that 17-AAG inhibited Akt activation at non-toxic doses and had anti-tumor activity in breast cancer xenografts.
Ansamycins induce the proteasomal degradation of several transmembrane tyrosine kinases, including members of the HER family (Mimnaugh et al., 1996; Xu et al., 2000). We determined whether the expression or activity of PI3 kinase or Akt was also directly affected. 17-AAG had no effect on the activity of either enzyme in vitro or on the intracellular expression of the p85 regulatory or p110 catalytic subunit of PI3 kinase (Figure 1c, data not shown (DNS)).
17-AAG caused a decrease in Akt protein expression in all cell lines examined (Figure 1a, DNS). The effect was detected by 12 h after drug addition and levels were reduced by 80% at 24 h. In most cells, the level of the phosphorylated, active form of Akt fell in parallel with that of the total Akt protein. The data suggest that inhibition of Akt expression by 17-AAG may contribute to its cellular effects.
In addition, in breast cancer cell lines with elevated expression of HER2 (SKBr-3 and BT-474), 17-AAG caused a rapid fall in Akt phosphorylation on serine 473 prior to any decline in Akt protein expression (Figure 1b). Phosphorylation of Akt on threonine 308 was undetectable by Western blot analysis in these cells. Akt phosphorylation and in vitro protein kinase activity fell in parallel beginning 1 h after drug addition and were undetectable by 1.5 h (Figure 1b). The concentration range required for inhibition of activation is 2 – 20 nm and levels were reduced to 30% of controls with 10 nm 17-AAG (Figure 1d).
Akt kinase has been shown to phosphorylate several key substrates that regulate protein translation, apoptosis and cellular proliferation (Marte and Downward, 1997; Vanhaesebroeck and Alessi, 2000). Phosphorylation of two of these substrates, glycogen synthase kinase-3 (GSK-3) and eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1), can be demonstrated in SKBr-3 cells (Figure 2a). 17-AAG caused dephosphorylation of these proteins at concentrations and times associated with inhibition of Akt activation. Akt has been shown to regulate D-cyclin translation and turnover (Diehl et al., 1998; Muise-Helmericks et al., 1998). In cells exposed to drug, cyclin D1 and D3 expression fell in parallel with Akt activity and dephosphorylation of Akt substrates (Figure 2b).
Akt kinase associates with the plasma membrane by binding to phosphatidyl inositol-3′ phosphates via its pleckstrin homology (PH) domain (Franke et al., 1995). At the membrane, Akt kinase is activated as a result of the phosphorylation of Thr-308 and Ser-473 by 3-phosphoinositide-dependent protein kinase 1 (PDK1) and perhaps other kinases (Alessi et al., 1997). Thus, Akt activation is dependent on PI3 kinase activity.
Ansamycins did not affect the expression of PDK1 (Figure 1c). Furthermore, drug treatment did not affect phosphorylation of PDK1 at Ser-241, a site essential for PDK1 activity (Figure 1c) (Casamayor et al., 1999). As previously described, they do induce the degradation of HER2 (Mimnaugh et al., 1996; Xu et al., 2000); expression was reduced by 90% after 2 h of exposure to as little as 10 nm 17-AAG (Figure 3a). We used 50 nm 17-AAG to ensure maximum inhibition. There was concomitant loss of tyrosine phosphorylation of HER2 (Figure 3a). Activated HER2 forms hetero- dimers with other members of the HER-kinase family and transphosphorylates its dimerization partner (Wallasch et al., 1995). One of these, HER3, has no endogenous tyrosine kinase activity, but contains multiple tyrosines that when phosphorylated are docking sites for PI3 kinase (Hellyer et al., 1998).
After 4 h of treatment, 17-AAG had no effect on HER3 expression, although HER3 levels do decline with longer exposure to drug (Figure 3a and DNS). However, association of HER3 with HER2 declined by 60% after 1 h and 90% after 4 h of treatment. Phosphorylation of HER3 declined in parallel (Figure 3a). Decreased phosphorylation of HER3 led to a coordinate decrease in its binding to the p85 regulatory subunit of PI3 kinase and in its association with PI3 kinase activity (50 – 65% decline at 2 h, 95% decline at 4 h, Figure 3a,b).
One hour pretreatment with lactacystin, a proteasome inhibitor, abrogated the effect of 17-AAG on HER2 protein (Figure 3c). Lactacystin is a specific irreversible inhibitor of the 20S proteasome that has been shown to block ansamycin depletion of HER2 protein (Fenteany et al., 1995; Mimnaugh et al., 1996). Treatment with lactacystin alone had no measurable effects on any of the proteins being studied (DNS). However, in the presence of lactacystin, 17-AAG still inhibited HER2 phosphorylation and caused decreased binding of HER3 to the p85 subunit of PI3 kinase and inhibition of Akt activation (Figure 3c). Additionally, 17-AAG induced reductions in cyclin D1 and D3 levels were not abrogated by lactacystin (Figure 3c, DNS). Thus, inhibition of HER2 activity by 17-AAG is coupled to but not dependent upon HER2 degradation.
17-AAG reduced HER2 expression, the association of HER3 with HER2, and HER3 phosphorylation and association with active PI3 kinase at concentrations and times correlated with loss of Akt activity. To test whether 17-AAG inhibited Akt activation by preventing the recruitment of PI3 kinase to the membrane by HER3, a p110 catalytic subunit of PI3 kinase engineered to bind the membrane was introduced into cells. In this protein the C-terminal of p110 is fused to the CAAX box (CVLS) of H-Ras, allowing farnesylation of p110 and its constitutive binding to the membrane in the absence of docking to receptor (Wennstrom and Downward, 1999).
Expression of p110-CAAX in SKBr-3 cells did not alter induction of degradation of HER2 by 17-AAG. In transfected cells, Akt activity was elevated fourfold and D-cyclin expression was increased threefold (DNS)(Muise-Helmericks et al., 1998). However, 17-AAG inhibited neither Akt phosphorylation nor kinase activity in the transfectants (Figure 4a). Moreover, in these cells, 17-AAG did not cause dephosphorylation of the Akt substrates 4E-BP1 and GSK-3 (Figure 2a). LY294002, a direct inhibitor of PI3 kinase, did inhibit Akt kinase in control and transfected cells (Figure 4b). Thus, inhibition of Akt activity by 17-AAG is a result of loss of PI3 kinase docking to HER3.
Ansamycins cause an RB-dependent G1 arrest associated with decreased D-cyclin expression (Srethapakdi et al., 2000). The sensitivity of breast cancer cells with high HER2 to 17-AAG is correlated with a rapid decrease in cyclin D1 and D3 expression (P Munster, unpublished data and Figure 2b). 17-AAG and LY294002 both inhibited cyclin D1 and D3 expression in SKBr-3 cells (Figure 2b). We were unsuccessful at detecting cyclin D2 in these cells. Additionally, 17-AAG inhibited cdk4 expression (DNS). The effect of 17-AAG on D-cyclins, but not cdk4, was abrogated in the p110-CAAX transfectants (Figure 2b). However, LY294002 potently inhibited Akt activity and D-cyclin expression in these cells (Figures 2b and and4b).4b). Thus, cyclin D1 and D3 expression is dependent on Akt activation by PI3 kinase in SKBr-3 cells. 17-AAG inhibited D-cyclin expression by downregulating this pathway upstream of PI3 kinase.
The potent inhibition of Akt activation by 17-AAG suggested that it might be a useful therapeutic agent in breast cancers with high levels of HER2 expression. We sought to determine whether inhibition of this pathway could be achieved in vivo at non-toxic doses of the drug. Unlike SKBr-3, BT-474 cells are tumorigenic when injected into nude mice and therefore we chose this model to study the in vivo effects of 17-AAG. BT-474 breast cancer cells overexpress HER2 and responded to 17-AAG in tissue culture in a fashion similar to SKBr-3 (DNS).
In mice, the maximally tolerated dose (MTD) of 17-AAG given daily for 5 days ranged from 75 – 125 mg/kg. Doses exceeding the MTD were associated with weight loss, elevated liver transaminase levels, anaemia and death. Mice treated with 17-AAG 75 mg/kg × 5 consecutive days with a second cycle repeated 2 weeks later demonstrated no gross toxicity or progressive weight loss. At this dose level, treatment resulted in a dose-dependent inhibition of the growth of the tumor xenografts (Figure 5a, DNS). A maximum mean tumor regression of 58% was noted on day 25, the final day of cycle 2.
The effect of treatment on Akt activity was determined after a single dose of 50 mg/kg 17-AAG. Mice were sacrificed pretreatment and at various times after drug injection. HER2 expression fell 50% by 4 h after treatment and 90% after 10 h. Two hours after drug exposure, phosphorylated Akt was reduced by 90%. Levels of both HER2 and phosphorylated Akt remained depressed 24 h after drug administration. Under these conditions, there was no significant change in Akt or p85 expression nor was any toxicity noted (Figure 5b, DNS). In contrast, MCF-7 xenografts treated with a single dose of 17-AAG (50 mg/kg) demonstrated no comparable rapid loss of P-Akt (DNS). 17-AAG at doses tolerable to the host had anti-tumor activity and induced HER2 degradation and Akt inactivation in breast tumor xenografts that overexpress HER2.
Akt kinase plays an important role in the pleiotropic control of cell growth, integrating pathways that regulate proliferation, apoptosis and metabolism (Marte and Downward, 1997; Vanhaesebroeck and Alessi, 2000). Engagement of the PI3 kinase pathway activates Akt and is required for transformation by tyrosine kinase receptors (Bardelli et al., 1999; Penuel and Martin, 1999; Skorski et al., 1997). The PTEN tumor suppressor gene encodes a lipid phosphatase that dephosphorylates phosphatidyl inositol-3′ phosphates and inhibits Akt (Stambolic et al., 1998). Mutational inactivation of PTEN is a common event in human cancer and results in high levels of Akt activity (Li et al., 1997; Steck et al., 1997). Thus, activation of Akt is a potent oncogenic stimulus and often occurs in human cancers. Functional inhibitors of Akt might be expected to inhibit tumor cell growth and increase their sensitivity to stimuli that induce apoptosis (Page et al., 2000).
We showed here that the ansamycin antibiotic 17-AAG inhibits Akt function in tumor cells in a complex manner. Ansamycins cause the degradation of a subset of proteins that require Hsp90 for conformational maturation (Czar et al., 1997; Mimnaugh et al., 1996; Pratt, 1998; Schulte et al., 1996; Stancato et al., 1997; Webb et al., 2000; Whitesell and Cook, 1996). In this study, the drug caused a reduction in Akt protein levels after 12 – 24 h of treatment. This was accompanied by a parallel decline in Akt activity. Akt may be a direct target of ansamycins. Hsp90 has recently been reported to associate with Akt and regulate its activity (Sato et al., 2000). Our preliminary data suggests that Akt associates with Hsp90 and 17-AAG results in the proteasomal degradation of Akt.
In most of the cells that were tested, ansamycins depressed cellular Akt activity by reducing its expression. However, in breast cancer cells that express high levels of HER2, 17-AAG also caused rapid inhibition of Akt kinase activity. Inhibition was secondary to down-regulation of HER2-dependent signaling. Ansamycins induce the degradation of mature membrane bound HER2 by an Hsp90-dependent process, but the underlying mechanism of this is not completely defined (Xu et al., 2000). In these cells, active HER2 forms heterodimers with HER3, a kinase defective protein that is phosphorylated by HER2. Tyrosine phosphorylated HER3 binds to PI3 kinase and brings it to the membrane. Degradation of HER2 in cells exposed to 17-AAG led to HER3 dephosphorylation, loss of its association with PI3 kinase, and a rapid decline in Akt activity. As a consequence, the phosphorylation of Akt substrates declined, cyclin D expression fell and the cells were sensitized to agents that induce apoptosis. We have shown that 17-AAG causes modest apoptosis and sensitizes cells to the apoptotic effect of taxanes (Munster et al., 2001). Introduction into cells of a p110 catalytic subunit of PI3 kinase that does not require docking to a receptor in order to bind the membrane prevented these effects of 17-AAG (unpublished data).
Lactacystin is a specific irreversible inhibitor of the 20S proteasome (Fenteany et al., 1995). This inhibitor abrogated the effect of 17-AAG on HER2 protein expression but failed to prevent the loss of HER2 and Akt phosphorylation, and D-cyclin expression. In cells pretreated with lactacystin, 17-AAG treatment resulted in a loss of HER2 from the plasma membrane and an accumulation of HER2 within cytoplasmic vesicles (DNS). It is likely that HER2 loss following 17-AAG treatment involves its ubiquitination and targeting to the proteasome (Mimnaugh et al., 1996). These data suggest ubiquitinated HER2 is functionally inactive prior to proteasomal degradation.
These results have several important implications. Amplification and overexpression of the HER2 gene occurs in 30% of breast tumors and a significant number of other cancers (Slamon et al., 1987, 1989; Tal et al., 1988). Overexpression is associated with an aggressive clinical phenotype and an antibody directed against the extracellular domain of HER2 has anti-tumor activity in patients (Baselga et al., 1996; Pegram et al., 1998). Our data suggest that in breast cancer cells with elevated HER2 and HER3 expression, Akt kinase is active and required for efficient expression of cyclin D1 and D3. Thus, in these tumors HER2 may function both to accelerate G1 progression by deregulating D-cyclin expression and to suppress apoptosis via other Akt-dependent pathways.
Inhibition of Akt function by 17-AAG may be responsible for many of its biologic effects. Despite their spectrum of targets, the cellular effects of ansamycins are quite selective. They cause RB-dependent G1 arrest with subsequent apoptosis (Srethapakdi et al., 2000). The RB-dependence of the arrest suggests that ansamycins selectively downregulate pathways required for cyclin D-cdk activation. Hsp90 is involved in the stabilization of cdk4, a protein targeted by the ansamycins (Stepanova et al., 1996). Additionally, the drug causes loss in D-cyclin expression by downregulating a PI3 kinase, Akt dependent pathway (Muise-Helmericks et al., 1998). Akt regulates both the translation and stability of cyclin D-protein (Diehl et al., 1998; Muise-Helmericks et al., 1998). The work described here suggests that inhibition of signaling pathways that activate Akt together with direct effects on its expression may account for both the anti-proliferative and pro-apoptotic effects of ansamycins.
The anti-HER2 antibody Herceptin has anti-tumor activity in patients with breast cancer that is enhanced in combination with cytotoxic agents (Gilewski et al., 2000; Pegram et al., 1998, 1999). A drug that induces the potent degradation of HER2 together with a rapid loss in Akt activity could have significant therapeutic utility, but may be toxic in vivo. In this report, we show that 17-AAG caused downregulation of HER2 expression and Akt inactivation and had anti-tumor activity at doses that were tolerable to the animal. Furthermore, in ongoing phase 1 clinical trials, peak serum levels of 2 µm 17-AAG have been obtained in the absence of significant toxicity (H Scher, N Rosen, unpublished data). In tissue culture models, less than 10 nm 17-AAG was required for marked inhibition of Akt kinase in breast cancer cells that overexpress HER2. Although one cannot presume that serum drug levels reflect intratumoral levels, these data suggest that inhibiting Akt kinase in tumors by 17-AAG is feasible and thus represents a novel strategy for cancer treatment.
17-AAG (NSC 330507, NCI, Bethesda, Maryland, USA), LY294002 (Biomol, Plymouth Meeting, Pennsylvania, USA), and lactacystin (Sigma, St Louis, Missouri, USA) were dissolved in 100% DMSO. Lipids (Sigma) were stored in chloroform and sealed under nitrogen.
The human cancer lines MCF-7, MDA-MB-468 (MDA-468), SKBr-3, BT-474 (American Type Culture Collection, Manassas, Virginia, USA) were maintained in 1 : 1 mixture of DME:F12 supplemented with 2 mm glutamine, 50 units/ml penicillin, 50 units/ml streptomycin, and 10% heat inactivated fetal bovine serum (Gemini Bioproducts, Calabasa, California, USA) and incubated at 37°C in 5% CO2.
Polyclonal antibodies: Akt, P-Akt (Ser-473), P-4E-BP1 (Ser-65), P-GSK-3 (Ser-21/9), P-PDK1 (Ser-241) (Cell Signaling, Beverly, Massachusetts, USA); p85, p110α, PDK1, GSK-3 (Upstate Biotechnology, Lake Placid, New York, USA); Cyclin-D1 (m-20), 4E-BP1 (r-113), HER2 (c-18), HER3 (c- 17), P-HER2 (Santa Cruz Biotechnology, Santa Cruz, California, USA). Monoclonal antibodies: PY99 (Santa Cruz Biotechnology). HER3 (Ab-7) (Neomarkers, Fremont, California, USA) was used for immunoprecipitation.
Cells were exposed to 17-AAG or DMSO vehicle. Cells were lysed in NP-40 buffer (50 mm Tris, pH 7.5, 1% Nonidet P-40, 150 mm NaCl, 2.5 mm Na3VO4, 10mm phenylmethylsulfonyl fluoride, and 10 µm each leupeptin, aprotinin, and soybean trypsin inhibitor) and cleared by centrifugation. Protein concentration was determined by using BCA reagent (Pierce Chemical Co., Rockford, Illinois, USA). Immunoprecipitations were performed using Protein-G Sepharose (Amersham, Piscataway, New Jersey, USA). Samples were separated by 7 – 10% SDS – PAGE, transferred to nitrocellulose, immunoblotted and detected by chemiluminescence using the ECL detection reagents (Amersham).
Kinase activity was assayed using New England Biolabs Akt Kinase Kit. Akt was immunoprecipitated, washed twice with lysis buffer, then twice with kinase buffer (25 mm Tris pH 7.5, 5 mm beta-glycerolphosphate, 2 mm DTT, 0.1 mm Na3VO4, 10 mm MgCl2). Two hundred µm ATP and 1 µg substrate (paramyosin fused to a GSK-3 crosstide) were added and assays were performed at 30°C for 30 min. Reaction mixtures were separated by 10% SDS –PAGE and the P-GSK3 reaction product was detected by immunoblotting.
Cells were lysed in 137 mm NaCl, 20 mm Tris pH 7.5, 1 mm MgCl2, 10% v/v glycerol, 1% v/v Triton X, 10 mm phenylmethylsulfonyl fluoride, and 10 µm each leupeptin, aprotinin, and soybean trypsin inhibitor. HER3 immunoprecipitation complexes were washed three times with 1% Triton X-100 in PBS, twice with 0.5 m LiCl, 0.1 m Tris pH 7.5, and twice with 10 mm Tris pH 7.5, 100 mm NaCl, 1 mm EDTA. Complexes were incubated for 20 min at room temperature with 1 µm ATP, 5 µCi ATP (γP32), and a 0.5 µm/ml 1 : 1 lipid mix of phosphatidylinositol and phosphatidylserine in 10 mm HEPES pH 7.0, 1 mm EGTA. The reaction was quenched by 1 m HCl and lipids were extracted with 1 : 1 CHCl3 : CH3OH. The organic layer was separated by thin layer chromatography eluted with 50 : 15 : 35 1-propanol : methanol : glacial acetic acid and detected by autoradiography.
The activated PI3k construct was provided by J Downward. The carboxy-terminal farnesylation signal from H-Ras was fused to p110 (myc-tagged p110-CAAX in pSG5 vector) (Wennstrom and Downward, 1999). Two million cells were transfected with 2 µg of DNA and 10 µl Lipofectin reagent (Life Technologies, Rockville, Maryland, USA). Experiments were performed 24 h after transfection.
Six-week-old athymic BALB/c female mice (NCI-Frederick Cancer Center) were maintained in pressurized ventilated cages. Experiments were carried out under an IACUC approved protocol and institutional guidelines for the proper, humane use of animals in research were followed. Prior to tumor cell inoculation, 0.72 mg SR 17B-estradiol pellets (Innovative Research of America, Sarasota, Florida, USA) were placed subcutaneously into the right flank. 1 × 107 BT-474 cells were mixed 1 : 1 with Matrigel (Collaborative Research, Bedford, Massachusetts, USA) and injected subcutaneously. Mice were randomized to treatment or control groups and treated with 17-AAG (75 mg/kg) or the egg-phospholipid (EPL) vehicle alone in cycles of 5 consecutive days. Treatment cycles were repeated 2 weeks later. Mice were weighed and tumors were measured with vernier calipers. Tumor volumes were calculated with the formula: π/6 × larger diameter × (smaller diameter)2. To analyse cellular markers, mice with established tumors were treated with a single dose of 17-AAG 50 mg/kg i.p. Mice were sacrificed pre-treatment and 1, 2, 4, 10 and 24 h post-treatment. For immunoblotting, tumor tissue was homogenized in 2% SDS lysis buffer (pH 7.4).
We thank J Downward and S Wennstrom for providing the p110-CAAX construct. We thank M Moasser for his helpful discussions and technical assistance. This work was in part supported by the NCI Breast SPORE program grant P50CA68425, grant number CA0951 from the National Cancer Institute and the generous support of the Arlene Taub Foundation.