The Akt family of serine/threonine protein kinases are putative oncogenes that enhance the survival of various cell types.1, 2 The 3 closely related Akt isoforms induce cell proliferation and soft agar growth, and disrupt the acinar architecture of human mammary cells, but do not induce transformation by themselves.3, 4, 5 Akt activation requires 2 critical phosphorylation steps at Thr308 and Ser473. The phosphorylation is induced by ligand-mediated activation of growth factor receptors and receptor tyrosine kinases. Insulin-like growth factor I (IGF-I), insulin and the epidermal growth factor are ligands that activate Akt through the receptor-mediated pathway.1, 6 The phosphorylation of membrane-bound Akt establishes the full Akt kinase activity and induces Akt translocation into the cytosol and nucleus.7, 8, 9, 10 Activated Akt phosphorylates a number of downstream targets such as BAD, caspase-9, p21, p27, GSK3β, Mdm2, mTOR, IKKα and others.1, 11, 12 The phosphorylation of BAD and caspase-9 blocks the apoptotic activity of the 2 proteins and raises the threshold for apoptosis.13, 14, 15 Thus, tumors that show a high phosphorylation of BAD and caspase-9 may have a poor response to treatment.
Several key pathways in breast cancer biology activate the Akt pathway.4, 16, 17, 18 Akt expression is upregulated in breast tumors and the phosphorylation of Akt is associated with poor disease-free survival.19, 20, 21 We pursued the hypothesis that a proinflammatory tumor environment increases Akt phosphorylation in breast tumors. It has previously been observed that cyclooxygenase-2 (COX2)-specific inhibitors disrupt Akt signaling in breast cancer cells.22 We also examined the relationship between serum IGF-I and Akt phosphorylation in breast tumors. Serum IGF-I is a breast cancer risk factor that may enhance tissue inflammation.23, 24, 25, 26 We examined immunohistochemically the phosphorylation of Akt, BAD and caspase-9 in relation to nitric oxide synthase-2 (NOS2) and COX2 expression, tumor phagocyte density, and serum IGF-I. We also assessed the relationship between serum IGF-I and tissue inflammation, and examined possible interactions between these markers and the tumor Akt phosphorylation status in breast cancer survival.
Material and methods
Collection of tumor specimens and survival information
Paraffin-embedded surgical biopsy specimens, blood samples and survival information were obtained from 248 breast cancer cases at the University of Maryland Medical Center, Baltimore Veterans Affairs Medical Center, Union Memorial Hospital, Mercy Medical Center and Sinai Hospital in Baltimore, Maryland. The women were recruited between February of 1993 and August of 2003.27 Cases were eligible if they resided in the greater Baltimore area at the time of recruitment, were of African-American or Caucasian descent by self-report, had pathologically confirmed breast cancer, were diagnosed with breast cancer within the last 6 months prior to recruitment and had no previous history of breast cancer. Patients were excluded if they were HIV, HCV or HBV carriers, were IV-drug users, were institutionalized, or were physically or mentally unable to sign consent and complete the questionnaire. Of the eligible patients that were identified through surgery lists, 83% participated in the study. The subjects signed a consent form and completed an interviewer-administered questionnaire. The questionnaire evaluated the medical, reproductive, family and occupational history of the subjects. Additional information to determine body mass index (BMI), tumor size, node status, disease stage, treatment and survival was obtained from medical records and pathology reports, State of Maryland records, the Social Security Death Index and the National Death Index. The Institutional Review Boards at the participating institutions approved the protocol.
A pathologist reviewed H&E-stained sections of each tumor specimen, to confirm the presence of tumor and the histology. Disease staging was performed according to the tumor-node-metastasis (TNM) system (AJCC/UICC).
Formalin-fixed and paraffin-embedded 5-μm slides were deparaffinized and placed into citrate buffer for antigen retrieval. Slides were microwaved and rinsed in phosphate-buffered saline (PBS) buffer. Endogenous peroxidase was blocked using the DakoCytomation Envision System-HRP blocking buffer, according to the manufacturer's protocol (DakoCytomation, Carpinteria, CA). Protein expression and phosphorylation were evaluated using the following primary antibodies: 1:25 diluted rabbit polyclonal antibody (no. 9277; Cell Signaling Technology, Beverly, MA) for phosphorylated Akt (Ser473); 1:80 diluted monoclonal antibody 244F9 (no. 4056; Cell Signaling Technology) for phosphorylated Akt (Thr308); 1:100 diluted rabbit polyclonal antibody (no. 9295; Cell Signaling Technology) for phosphorylated Bad (Ser136); 1:250 diluted rabbit polyclonal antibody (no. SC-11755; Santa Cruz Biotechnology, Santa Cruz, CA) for phosphorylated caspase-9 (Ser196); 1:50 diluted monoclonal antibody (no. 610204; BD Biosciences/Transduction Laboratories, San Diego, CA) for COX2; 1:250 diluted monoclonal antibody (no. 610328; BD Biosciences/Transduction Laboratories) for NOS2; 1:100 diluted monoclonal DO-7 antibody (DakoCytomation) for p53; 1:100 diluted rabbit polyclonal antibody (DakoCytomation) for c-erbB-2 (HER2/neu); ready-to-use monoclonal Ab-3 antibody (Lab Vision, Fremont, CA) for CD68; and ready-to-use monoclonal (Clone 6F11) antibody (Ventana Medical Systems, Tucson, AZ) for the estrogen receptor (ER). After an overnight incubation with the primary antibody at 4°C, slides were washed in 1× PBS and incubated with a corresponding HRP-labeled secondary antibody, using the DakoCytomation Envision System reagents. Slides were washed after 30-min incubation at room temperature, were stained with DAB and counterstained with Methyl Green. Staining specificity was determined with negative and positive control slides and, if available, with blocking peptides that were purchased from the manufacturer. Phosphospecific peptides were available for phosphorylated Akt (pAkt) at Thr308 and Ser473, and phosphorylated caspase-9 at Ser196. To minimize staining variability, one person performed the immunostaining using a standard protocol. Images of stained sections were taken with the Olympus DP70 Digital Camera System (Olympus, Melville, NY).
Evaluation of the immunohistochemistry data
A combined score of intensity and distribution was used to categorize the immunohistochemical (IHC) staining for protein expression and phosphorylation, with the exception of the p53 and ER IHC. Intensity received a score of 0–3 if the staining was negative, weak, moderate or strong. The distribution received a score of 0–4 if the staining distribution was <10% positive cells, 10–30%, >30–50%, >50–80% and >80%. A sum score was then divided into 4 groups as follows: (i) negative = 0–1, (ii) weak = 2–3, (iii) moderate = 4–5 and (iv) strong = 6–7, according to standard protocols for IHC analysis.28, 29 The ER and p53 status were scored negative/positive. p53 expression was scored positive if >10% of the tumor cells expressed nuclear p53. The ER status was determined according to the reference range set by the ChromaVision® ACIS® assisted quantitative image analysis software (Clarient Diagnostic Services, Irvine, CA). The number of phagocytes in the tumor specimens was determined by counting the number of CD68-positive monocytes per 250× field in 3 representative fields.
IGF-I, IGF-II and IGFBP3 serum concentrations
Serum samples were available for 104 cases. IGF-I, IGF-II and IGFBP3 serum concentrations were determined as the average of duplicate measurements, using the Active® IGF-I, IGF-II and IGFBP3 ELISA kits, according to the manufacturer's protocol (Diagnostic Systems Laboratories, Webster, TX). The IGF-I serum concentration was also determined in the Department of Laboratory Medicine, NIH, using the Quest Diagnostics Nichols Institute IGF-I chemiluminescent assay (Nichols Institute Diagnostics, San Clemente, CA). The assays at the 2 laboratories produced highly correlative data (r2 = 0.98, p < 0.0001).
Coculture of human breast cancer cells with ANA-1 mouse macrophages
MCF-7 cells were seeded at 2 × 105 cells in 6-well plates and grown in RPMI medium with 4.5 g/l glucose, 2 mM glutamine and 10% FBS. The cells were grown to 80% confluence and serum-starved overnight. ANA-1 cells were cultured in high glucose DMEM + 10% FBS and activated overnight with interferon-γ (10 ng/ml) followed by LPS (10 ng/ml) for 4 hr. ANA-1 cells were harvested and added to the serum-starved MCF-7 cells at a 1:1 ratio. The coculture was stopped at 1, 2 and 4 hr when the cell culture medium was removed, and the MCF-7 cells were washed twice with cold PBS, pH = 7.4, and solubilized with cold lysis buffer (50 mM Tris/HCl, pH = 7.5, 5 mM EDTA, 150 mM NaCl, 1% Triton-X-100 and a protease inhibitor mix). Most of the ANA-1 cells are removed from the coculture during the washing steps with PBS. The coculture experiments with MDA-MB-231 cells were conducted according to a previously published protocol30 with the following modifications: MDA-MB-231 cells were separated from the ANA-1 cells using the MACS Separator System (Miltenyi Biotec, Auburn, CA). Briefly, following coculture, cells were trypsinized, counted and washed 2× with PBS. Cells were resuspended in 80 μl separation buffer (PBS, 2 mM EDTA, 0.5% BSA) per 1 × 107 cells plus 20 μl anti-mouse CD45 beads. The mixture was incubated at 4°C for 15 min. The MACS Separator column was placed into a miniseparator, and 0.5 ml buffer was added to the column to balance, and to allow for flow through. The cell mixture was washed with 1 ml separation buffer and centrifuged. The pellet was resuspended with 80 μl separation buffer per 1 × 107 cells and added to the column. Cells flowing through were collected, and the column was washed 3 times with separation buffer. The cell suspension was centrifuged, and the supernatant discarded. The pellet of separated, purified MDA-MB-231 cells was stored at −80°C until analysis.
Treatment of human breast cancer cells with SPER/NO, prostaglandin E2 and okadaic acid
MCF-7 cells and MDA-MB-231 cells were grown to 70% confluency in 100-mm plates in RPMI medium with 2 g/l glucose, 2 mM glutamine and 10% FBS, and DMEM medium with 4.5 g/l glucose, 2 mM glutamine, 1 mM sodium pyruvate and 10% FBS, respectively. Cells were treated with either 25 μM SPER/NO (Alexis Biochemicals, San Diego, CA) or 5 μM prostaglandin E2, (Sigma-Aldrich, St. Louis, MO). MCF-7 cells were treated with 10 nM okadaic acid and MDA-MB-231 cells with 100 nM okadaic acid (Calbiochem, San Diego, CA). A third cell line, MDA-MB-157, was only treated with 25 μM SPER/NO.
Western blot analysis
Protein concentrations were determined with the Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA). Western blot analysis was performed according to standard procedures, and 50 μg of total protein was loaded per lane. The protein bands were visualized using the SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL). The following antibodies were used to detect the membrane-bound proteins: rabbit polyclonal anti-pAkt Ser473 from Cell Signaling, 1:500 (no. 9271); rabbit polyclonal anti-pAkt Thr308 from Cell Signaling, 1:250 (no. 9275); rabbit polyclonal anti-Akt from Cell Signaling, 1:2,000 (no. 9272); rabbit polyclonal anti-pGSK3β Ser9 from Cell Signaling, 1:500 (no. 9336); rabbit polyclonal anti-NOS2 from Cayman Chemical, Ann Arbor, MI, 1:400 (no. 160862); monoclonal anti-COX2 from BD Biosciences/Transduction Laboratories, 1:250 (no. 610204).
Stata 7.0 (Stata, College Station, TX) and SAS 8.0 (SAS Institute, Cary, NC) statistical softwares were used for data analysis. All statistical tests were 2-sided, and an association was considered statistically significant with p < 0.05. The Spearman correlation coefficient was calculated for correlation analyses. The χ2 test and multivariate logistic regression were performed to analyze dichotomized data and to calculate adjusted odds ratios, respectively. To generate dichotomized data for IHC, scores were divided into moderate to strong versus negative to weak, or into strong versus negative-moderate. The phosphorylation status of Akt, BAD and caspase-9 was dichotomized into strong versus negative to moderate in the logistic and Cox regression analyses, because most tumors had a moderate to strong phosphorylation score for these proteins (Table I). Continuous data, such as serum protein concentrations, BMI or CD68 count, were dichotomized using the median value as the cutoff. Survival analysis was conducted in univariate and multivariate modes. The Kaplan–Meier method and the log-rank test were used for univariate survival analysis. The Cox Proportional-Hazards regression was used for multivariate survival analysis to calculate adjusted hazard ratios. Proportional-hazards assumptions were verified by log–log plots and with the non-zero slope test of the scaled Schoenfeld residuals. A statistical test for interaction was performed to determine whether the effect of Akt phosphorylation on breast cancer survival is modified by other factors. In this test, Akt phosphorylation was coded as zero (negative to moderate IHC) and one (strong IHC). Survival was determined for the time period from the date of hospital admission to the last completed search for death entries in the Social Security Death Index (August 18, 2004). The mean follow-up time for breast cancer survival was 55 months (range: 12–140 months). We obtained death certificates of the deceased cases, and censored all causes of death that were not related to breast cancer, such as accidents, violent crimes, stroke, heart attack and liver cirrhosis, in our analysis.
Table I. Summary of Tumor IHC and Serum Marker Analysis
We examined immunohistochemically the phosphorylation status of Akt (Akt1-3) at Thr308 and Ser473, BAD at Ser136, caspase-9 at Ser196, and determined the density of CD68-positive tumor phagocytes and the expression of NOS2 and COX2 in 248 surgical specimens of breast cancer (Table I). Phosphorylation of Akt, BAD and caspase-9 are markers of Akt pathway activation, while tumor phagocyte density, and expression of NOS2 and COX2 are indicators of a proinflammatory tumor environment. We also determined the tumor HER2/neu and p53 status, because both markers have previously been associated with Akt phosphorylation in breast cancer.19, 21 IGF-I, IGF-II and IGFBP3 serum concentrations were measured for the 104 cases with available serum samples (Table I). The demographic and clinicopathological data of the patients are shown in Table II.
Table II. Demographic and Clinicopathological Features of Cases1
Cases with missing information are not included. Values are number (percentage) unless otherwise indicated.
Premenopausal: still having menstruation; postmenopausal: 55 years or older, or stated change of life, or hysterectomy with both ovaries removed.
We found a moderate to strong phosphorylation of Akt, BAD and caspase-9 in most breast tumors (Table I). The immunostaining of pAkt is shown in Figure 1a. The staining for the Thr308-pAkt indicated a cytosolic distribution with additional nuclear localization. Phosphorylation at Ser473 showed a cytosolic distribution with a more intensive staining at the inner cell membrane. Phosphorylated BAD and caspase-9 had a granular cytosolic distribution, which is consistent with a cytosolic and mitochondrial localization of the phosphorylated proteins (Fig. 1b). Phosphorylated BAD was also detected in the nucleus of tumor cells in a subset of tumors. The intensity and distribution of Akt, BAD and caspase-9 phosphorylation correlated strongly between each other (Table III). NOS2 and COX2 were expressed in both the tumor epithelium and stromal cells (Fig. 1b). The expression of both NOS2 and COX2 was significantly associated with tumor phagocyte density (p = 0.016 for NOS2 and p = 0.01 for COX2; Spearman rank correlation). We did not observe a correlation between serum IGF-I and markers of inflammation. Neither the expression of NOS2 and COX2, nor the density of tumor phagocytes in breast tumors, were associated with serum IGF-I (data not shown).
Table III. Phosphorylation of Akt, Bad and Caspase-9 in Relationship to Tumor and Serum Markers
Akt pathway activation in relationship to markers of inflammation
We examined the relationship between the expression of NOS2 and COX2, tumor phagocyte density, and Akt pathway activation (Table III). The Spearman rank correlation analysis showed very significant correlations between NOS2 and COX2 expression, and the phosphorylation of Akt, BAD, and caspase-9. A high expression of NOS2 and COX2 was associated with a strong phosphorylation of these proteins. We also found a correlation between HER2/neu overexpression and an increased phosphorylation of Akt, BAD and caspase-9 (Table III). The tumor phagocyte density and aberrant p53 expression were significantly associated only with the phosphorylation of Akt at Ser473, but not at Thr308. Other factors, such as ER, node, menopausal status, BMI, smoking, age, or race/ethnicity, were not associated with Akt pathway activation. We further corroborated the relationship between NOS2, COX2 and Akt pathway activation in a multivariate logistic regression analysis. NOS2 expression was an independent predictor of pAkt at Ser473 (Odds ratio (OR) = 2.9; 95% confidence interval (CI), 1.1–8.3), but not of pAkt at Thr308 (OR = 1.3; 95% CI, 0.5–4.0), in a multivariate logistic regression model that included COX2, HER2/neu, p53, IGF-I, tumor phagocytes, TNM stage and neoadjuvant therapy as covariates. COX2 expression predicted both phosphorylation at Thr308 (OR = 3.3; 95% CI, 1.2–9.4) and at Ser473 (OR = 3.0; 95% CI, 1.1–8.6) in the same multivariate analysis. The data suggest that NOS2 induces phosphorylation of Akt at Ser473, and COX2 of Akt at Thr308 and Ser473 in human breast tumors, independent of the other markers that were associated with Akt pathway activation in our study. The 2 proteins were also very significant predictors of phosphorylated caspase-9 at Ser196 (OR = 6.4; 95% CI, 1.6–25.5 and OR = 6.1; 95% CI, 1.9–20.6, respectively) in the multivariate analysis. HER2/neu overexpression, tumor phagocyte density and the tumor p53 status were not independently associated with Akt phosphorylation in our study, as judged by the logistic regression analysis (data not shown).
Akt pathway activation in relationship to IGF-I
We investigated the relationship between serum IGF-I and Akt activation in breast tumors, using both the serum IGF-I concentration and the ratio between IGF-I and IGFBP3 as indicators of IGF-I bioavailability. Both IGF-I and the IGF-I/IGFBP3 ratio showed very similar and statistically significant associations with the phosphorylation status of Akt and BAD in tumor tissues (Table III). The observed association with Akt phosphorylation at Thr308 was independent of other markers that were associated with protein phosphorylation in our study. The multivariate logistic regression analysis showed that serum IGF-I and the IGF-I/IGFBP3 ratio were predictors of Akt phosphorylation at Thr308 (OR = 5.2; 95% CI, 1.7–16.5 and OR = 2.9; 95% CI, 1.1–7.8, respectively), after adjustment for IGF-II, NOS2, COX2, HER2/neu, p53, tumor phagocytes, TNM stage and neoadjuvant therapy. The association with Akt phosphorylation at Ser473 was weaker and not statistically significant in the multivariate analysis. We also examined the correlation between serum IGF-II and Akt phosphorylation to discriminate between the effect of serum IGF-I and IGF-II with respect to Akt activation in breast tumors. Serum IGF-II was not significantly associated with Akt pathway activation in our study (Table III).
Phosphorylation of Akt in human breast cancer cells treated with nitric oxide and prostaglandin E2
We further investigated the relationship between NOS2 and COX2 activity and pAkt, and assessed whether the major metabolites of NOS2 and COX2 induce the phosphorylation of Akt at Thr308 and Ser473. We examined the effect of nitric oxide and prostaglandin E2 on Akt phosphorylation in human breast cancer cells. MCF-7 and MDA-MB-231 cells were treated with 25 μM SPER/NO, a nitric oxide donor, for 1–4 hr, (Fig. 2a). At this concentration, SPER/NO generates 100–400 nM nitric oxide concentrations in the cell culture medium. Both cell lines contained detectable amounts of pAkt at Ser473 at baseline. Nitric oxide increased Akt phosphorylation at Ser473 but did not lead to a detectable enhancement of the phosphorylation at Thr308 in the 2 cell lines. We also observed an induction of Akt phosphorylation at Ser473 by nitric oxide in a third breast cancer cell line, MDA-MB-157. Prostaglandin E2, the most active proinflammatory metabolite of COX2, stimulated Akt pathway activation as well (Fig. 2b). At a 5 μM concentration, prostaglandin E2 induced Akt phosphorylation at Ser473 and GSK-3β at Ser9, a known downstream target of pAkt, in both cell lines. The prostaglandin E2 response appeared to be stronger in MDA-MB-231 cells than in MCF-7 cells.
Treatment of human breast cancer cells with okadaic acid
To test whether Akt phosphorylation may lead to an induction of NOS2 and COX2 expression, we treated MCF-7 cells and MDA-MB-231 cells with 10 and 100 nM okadaic acid, respectively. In the initial titration experiment with 0, 10, 100, 300 nM okadaic acid, these 2 concentrations induced a similar phosphorylation of Akt at Ser473 in the 2 cell lines (data not shown). Okadaic acid is an inhibitor of the protein phosphatase 2A and leads to an accumulation of pAkt and activation of the Akt pathway.31, 32 We exposed cells for 48 hr time to okadaic acid, and examined the induction of NOS2 and COX2 by Western blot analysis (Fig. 2c). As expected, okadaic acid induced the activation of the Akt pathway, as shown by the phosphorylation of Akt at Ser473 and GSK3β at Ser9. We could not detect an induction of NOS2 in the 2 cell lines, but observed an induction of COX2 at 48 hr in MDA-MB-231 cells (Fig. 2c). The data suggest that Akt pathway activation in human breast cancer cells does not lead to an induction of NOS2. Akt activation, however, appears to induce the expression of COX2 in at least some breast cancer cells.
Coculture of human breast cancer cells with ANA-1 macrophages
We used a coculture experiment with human breast cancer cells and ANA-1 mouse macrophages to corroborate the association between tumor phagocytes and pAkt in breast tumors. We found that ANA-1 mouse macrophages induce phosphorylation of Akt in MCF-7 cells during a time course of 1–4 hr (Fig. 3a). The detected phosphorylation at Ser473 appeared to be much stronger than the phosphorylation at Thr308. The induction was not dependent on the activation status of the macrophages; untreated and INF-γ/LPS-stimulated macrophages induced pAkt to a similar extent. We repeated the coculture experiment with a second breast cancer cell line, MDA-MB-231. Again, we found that unstimulated ANA-1 cells induced a phosphorylation of Akt that appeared to be stronger at Ser473 than at Thr308 (Fig. 3b).
Akt pathway activation and breast cancer survival
We investigated the association between Akt pathway activation and overall survival, and focused on the inhibition of apoptosis as a predictor of outcome. We censored all causes of death that were not related to breast cancer. Hence, overall survival in our study meets the definition of breast cancer survival.
We first studied the association between the phosphorylation status of Akt, BAD and caspase-9, and the breast cancer outcome using Kaplan–Meier survival analysis (Fig. 4). The Kaplan–Meier method and the log-rank test did not show an association between the phosphorylation status of Akt, BAD and caspase-9 in breast tumors, and breast cancer survival. This did not change when we dichotomized the data at a different cutoff point, using the immunostaining score of moderate to strong, instead of strong, to define high protein phosphorylation. We further corroborated our observations with a multivariate analysis, using the Cox Proportional-Hazards Regression and adjustments for age at diagnosis, race, BMI, ER status, TNM stage and chemotherapy. In this analysis, neither phosphorylation of Akt, BAD or caspase-9 was significantly associated with breast cancer outcome.
Because we hypothesized that tissue inflammation may biologically interact with the Akt pathway in breast tumors and modify the effect of the Akt phosphorylation status on breast cancer survival, we stratified our survival analysis by tumor phagocyte density. We observed that the tumor phagocyte density modified the effect of Akt phosphorylation on breast cancer survival in our study population (Table IV). In tumors with an above median phagocyte count, a high pAkt was found to be associated with improved survival. This finding was statistically significant for the Thr308 phosphorylation and suggestive for the Ser473 phosphorylation. Because of this unexpected result, we further stratified by cancer treatment. The stratification indicated that the association could be related to neoadjuvant therapy (Table IV). We corroborated our finding with a statistical test for interaction. Consistent with the survival analysis, we found statistical interactions between Akt phosphorylation at Thr308 and Ser473, and tumor phagocyte density on breast cancer survival (pinteraction < 0.05, respectively). The interaction was particularly strong in the subgroup of patients who received neoadjuvant therapy (pinteraction = 0.001 and pinteraction = 0.01, respectively).
Table IV. Association Between Tumor Akt Phosphorylation and Overall Survival After stratification by Tumor Phagocyte Count
Cox Proportional-Hazards regression adjusted for age at diagnosis, race, BMI, estrogen receptor, TNM stage and chemotherapy.
Hazard ratio (HR) and 95% confidence interval (CI) comparing strong IHC versus negative-moderate IHC as the reference group.
HR with IHC entered as categorized data (1, 2, 3, 4 = negative, weak, moderate, strong). Values in parentheses indicate 95% confidence intervals.
< Median count
≥ Median count
< Median count
≥ Median count
We studied the phosphorylation of Akt, BAD and caspase-9 in 248 surgical breast tumors by IHC analysis and found that most breast tumors have moderate to high levels of pAkt, BAD and caspase-9. To our best knowledge, this is the first study that assessed the phosphorylation status of BAD and caspase-9 in human breast tumors. A deregulation of the PI3K pathway is the most common genetic alteration in breast cancer with 10–35% of breast tumors having a diminished PTEN expression, while another 25–30% may harbor a PIK3CA mutation.33, 34, 35 These genetic changes are thought to result in a constitutive activation of the Akt pathway. Our data are consistent with these findings by showing that a moderate to strong phosphorylation of Akt, BAD and caspase-9 is frequent in human breast tumors.
The intensity and distribution of Akt, BAD and caspase-9 phosphorylation correlated strongly between each other. The finding is an additional validation of the IHC results, because the phosphorylation of Akt at Ser473 and Thr308 are sequential steps that lead to kinase activation and phosphorylation of downstream targets, such as BAD and caspase-9.1 We found that Ser473-pAkt was partly membrane-bound and Thr308-pAkt had a predominantly cytosolic and nuclear localization. The observations are in agreement with the model that Ser473 phosphorylation precedes Thr308 phosphorylation, which leads to membrane detachment and nuclear translocation.9, 10 We observed a cytosolic and mitochondrial localization of phosphorylated BAD and caspase-9, and also a nuclear localization of phosphorylated BAD in a subset of the breast tumors. BAD and caspase-9 are both mitochondrial proteins that redistribute from the mitochondria to the cytosol upon stress or protein phosphorylation.13, 14, 15, 36 Thus, our IHC results are consistent with the reported localization of these phosphorylated proteins in cell culture.
Phosphorylation of Akt, BAD and caspase-9 is a sequence that activates a survival pathway and raises the threshold for apoptosis. The phosphorylation of BAD and caspase-9 inhibits their proapoptotic function.13, 14, 15 We observed that 3 markers of a proinflammatory environment, the expression of NOS2 and COX2, and the tumor phagocyte density, were associated with the phosphorylation of Akt, BAD and caspase-9 in breast tumors. The tumor phagocyte density was significantly associated with the phosphorylation of Akt at Ser473; however, the association was not independent of other tumor markers in the multivariate analysis. Phagocytes express NOS2 and can produce high amounts of nitric oxide.37 Our multivariate analysis indicated that NOS2 is a confounder of the relationship between tumor phagocytes and Akt phosphorylation, suggesting that the release of nitric oxide by tumor phagocytes may induce the phosphorylation of Akt at Ser473 in the tumor epithelium.
NOS2 and COX2 were both independently associated with Akt pathway activation in the multivariate analysis, after adjustment for potential confounders; however, NOS2 expression only predicted the Ser473 phosphorylation, while COX2 expression predicted both Thr308 and Ser473 phosphorylation of Akt. The data indicate that both enzymes activate the Akt pathway in a dominant manner in human breast tumors but NOS2 may primarily induce the phosphorylation of Ser473. Alternatively, pAkt may have led to the upregulation of NOS2 and COX2 in breast tumors. We found that the activation of the Akt pathway by the phosphatase 2A-inhibitor, okadaic acid,31, 32 induces COX2 expression in MDA-MB-231 human breast cancer cells. Okadaic acid did not stimulate the expression of NOS2 in this cell line or another human breast cancer cell line, MCF-7. Furthermore, it has previously been shown for COX2, but not NOS2, that Akt activation is involved in the transcriptional upregulation of the enzyme in human cells.38 Thus, pAkt may induce COX2 expression in human breast tumors. However, the expression of a COX2 transgene induces Akt phosphorylation in human cells,39 and several recent studies have shown that COX2-specific inhibitors block Akt phosphorylation and disrupt Akt signaling in various cancer cell lines,22, 40, 41 supporting the hypothesis that COX2 upregulation, as part of an inflammatory response, leads to Akt phosphorylation in human tumors. Additionally, the proinflammatory COX2 metabolite, prostaglandin E2, has been shown to induce Akt phosphorylation in colon cancer cells.42 We exposed breast cancer cells to prostaglandin E2 and detected the phosphorylation of Akt and GSK-3β, a downstream target of pAkt.1 These data indicate that COX2 and its major proinflammatory metabolite, prostaglandin E2, stimulate Akt pathway activation in human cancer cells.
We also extended our finding that NOS2 expression and phagocytes may induce pAkt in breast cancer cells, with a preference for Ser473. The nitric oxide donor, SPER/NO, induced a detectable phosphorylation only at Ser473 in MCF-7, MDA-MB-231 and MDA-MB-157 human breast carcinoma cells. ANA-1 mouse macrophages stimulated a strong phosphorylation at Ser473 in both MCF-7 and MDA-MB-231 cells, but an increase in the phosphorylation of Thr308 was either undetectable or weak in our coculture experiments. It was not shown previously that phagocytes and nitric oxide induce Akt phosphorylation in human cancer cells but DETA/NO, a nitric oxide donor closely related to SPER/NO, was found to induce Akt Ser473 phosphorylation in RINm5F cells.43
Serum IGF-I was found to be associated with an increased breast cancer risk in population-based studies.6, 23 Although epidemiological and experimental studies indicate that IGF-I activates cancer-related pathways, an association between IGF-I and breast cancer outcome has not been reported, and very few studies attempted to link serum IGF-I to pathway activation in human tissue specimens. We hypothesized that IGF-I may induce low-level tissue inflammation24, 25, 44 and activate the Akt pathway. We did not observe an association between serum IGF-I concentrations and markers of inflammation in breast tumors, but found that serum IGF-I and the IGF-I/IGFBP3 ratio, a measure for bioavailable IGF-I, were associated with Akt pathway activation and predicted Akt phosphorylation at Thr308 in human breast tumors. Our data indicate that breast cancer patients with a high serum concentration of bioavailable IGF-I, either because of a genetic predisposition or because of their lifestyle,6 are more likely to have a strong phosphorylation of Akt in their tumors than those with a low concentration of bioavailable IGF-I. We did not find that IGF-I independently predicted the phosphorylation of Akt at Ser473, although it was shown to induce Akt phosphorylation at both Thr308 and Ser473 in cell culture.10, 45 It is possible that other pathways have a more dominant effect on the phosphorylation of Ser473 in breast tumors. IGF-I induces a transcriptional profile in breast epithelial cells that is consistent with the induction of angiogenesis and cancer progression.46 There is strong evidence, from inhibitor and knockout studies, that the IGF-I receptor promotes adhesion, invasion, metastasis, survival and drug-resistance of breast cancer cells.47, 48, 49 Our finding from human breast tumors are in agreement with cell culture-based observations that IGF-I activates the Akt survival pathway in cancer cells.
Previous studies investigated the association between the PI3K pathway and breast cancer outcome. Several studies found an association between Akt Ser473 phosphorylation and disease-free survival and/or the response to therapy in breast cancer.19, 20, 21 Another study reported that the loss of PTEN predicts trastuzumab resistance in breast cancer patients.50 However, 2 large studies did not observe an association between breast cancer survival and either PI3CA mutations, PTEN loss, or Akt Ser473 phosphorylation.34, 51 We could not study disease-free survival. Instead, we examined overall survival that was censored for deaths not related to breast cancer to meet the definition of breast cancer-specific survival. We did not find a significant association between Akt, BAD and caspase-9 phosphorylation, and breast cancer outcome in the univariate and multivariate analyses. The data suggests that Akt pathway activation may not be associated with overall survival in breast cancer in an unstratified analysis.
Because we hypothesized a priori that inflammation may interact with the Akt pathway in breast cancer survival, we performed further analyses and studied whether tumor phagocyte density modifies the association between the Akt phosphorylation status and breast cancer survival. We found a statistical interaction between pAkt and tumor phagocyte count on breast cancer survival, and observed that Akt phosphorylation was associated with increased breast cancer survival if the tumor phagocyte count was high. This relationship was particularly strong in patients who had received neoadjuvant therapy. This is a preliminary finding. Because of the exploratory nature of our analysis, we cannot exclude the possibility that the observed interaction is a chance finding. Therefore, further confirmation in other studies is needed to determine whether a proinflammatory tumor environment modifies the association between the Akt phosphorylation status and breast cancer survival.
We used CD68 as a marker for phagocyte infiltration. CD68 is expressed in cells of the mononuclear phagocyte lineage and is thought to be the best panmacrophage marker.52 Tumor-associated macrophages can be both tumoricidal and protumor.53, 54 A high density of phagocytes in tumors may cause cytotoxicity, angiostasis and tumor regression.55 Thus, the high phagocyte infiltration after neoadjuvant therapy may not only lead to an increased phosphorylation of Akt in breast tumors, but also to a strong antitumor response. Our data suggest that, in this environment of excessive inflammation, Akt signaling can be detrimental to tumor progression. It has recently been shown that the PI3K pathway can augment p53 tumor suppressor activity in chemotherapy-treated cells.56 There is also new evidence that activation of Akt1 can suppress tumor invasion.57, 58, 59 The data indicate that Akt pathway activation may not always promote tumor progression. If Akt activation would promote survival but also inhibit tumor invasion, this could explain why Akt phosphorylation is associated with poor disease-free survival, but not overall survival, in human breast cancer.
In summary, we found that a proinflammatory environment and serum IGF-I are markers of Akt pathway activation in breast cancer. The expression of NOS2 and COX2, and a high tumor phagocyte density, were associated with increased Akt phosphorylation in human breast tumors. Additional experiments with human breast cancer cells corroborated these findings.
The authors thank Mr. Neil Caporaso for his help with the study design and Mrs. Melissa Miller for help with the statistical analysis. We also thank Mr. Raymond Jones, Ms. Audrey Salabes, Mr. Leoni Leondaridis, Mr. Glennwood Trivers, Mrs. Elise Bowman, and personnel at the University of Maryland and the Baltimore Veterans Administration, and the Surgery and Pathology Departments at the University of Maryland Medical Center, Baltimore Veterans Affairs Medical Center, Union Memorial Hospital, Mercy Medical Center and Sinai Hospital for their contributions.