|Home | About | Journals | Submit | Contact Us | Français|
Triple negative breast cancer (TNBC) has the highest mortality among all breast cancer types and lack of targeted therapy is a key factor contributing to its high mortality rate. In this study, we show that 8-bromo-cAMP, a cyclic adenosine monophosphate (cAMP) analog at high concentration (> 1 mM) selectively suppresses TNBC cell growth. However, commonly-used cAMP-elevating agents such as adenylyl cyclase activator forskolin and pan phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX) are ineffective. Inability of cAMP elevating agents to inhibit TNBC cell growth is due to rapid diminution of cellular cAMP through efflux and decomposition. By performing bioinformatics analyses with publically available gene expression datasets from breast cancer patients/established breast cancer cell lines and further validating using specific inhibitors/siRNAs, we reveal that multidrug resistance-associated protein 1/4 (MRP1/4) mediate rapid cAMP efflux while members PDE4 subfamily facilitate cAMP decomposition. When cAMP clearance is prevented by specific inhibitors, forskolin blocks TNBC's in vitro cell growth by arresting cell cycle at G1/S phase. Importantly, cocktail of forskolin, MRP inhibitor probenecid and PDE4 inhibitor rolipram suppresses TNBC in vivo tumor development. This study suggests that a TNBC-targeted therapeutic strategy can be developed by sustaining an elevated level of cAMP through simultaneously blocking its efflux and decomposition.
Triple-negative breast cancer (TNBC), a subset of breast cancer with the absence of estrogen and progesterone receptors (ERs and PRs) and lack of amplification of the human epidermal growth factor receptor 2 (HER2) gene, accounts for 15–20% of all breast cancer [1, 2]. Compared to hormone receptor-positive breast cancer, TNBC more commonly affects younger patients, has a higher prevalence in African-American women and often follows an aggressive clinical course including a high incidence of visceral and brain metastases [3, 4]. Although cytotoxic chemotherapy, the only approved therapy for TNBC, leads to an initial significant response rate, patients frequently suffer from relapses and subsequent mortality . Therefore, innovative therapeutic strategies targeting TNBC are urgently needed .
Agents that increase cellular cyclic AMP (cAMP) level have been found to suppress cell growth in various cell types by triggering apoptosis or cell-cycle arrest [7–9]. These agents include cAMP analogs [dibutyryl-cAMP, 8-bromo-cAMP (8-Br-cAMP) and 8-chloro-cAMP], adenylate cyclase activators and phosphodiesterase (PDE) inhibitors. Unfortunately, none of them has been recommended for cancer therapy because they display high toxicity at the dose that can effectively inhibit in vitro tumor cell growth or in vivo tumor development . We reason that understanding the cause that these agents elicit anti-tumor effect only at very high doses can help developing strategies in which cAMP-elevating agents can be utilized at reduced and nontoxic doses.
Cellular events led by cAMP are generally mediated through protein kinase A (PKA) and cAMP-regulated guanine nucleotide exchange factors . PKA-II is preferentially expressed in normal non-proliferating tissues and growth-arrested cells whereas PKA-I is overexpressed in cancer cells . Since cAMP analogs inhibit PKA-I expression while they promote the formation of PKA-II in cancer cells, the differential regulation of PKA isozymes by cAMP may be one of the explanations for cAMP's growth-suppressive activity [13, 14]. Recent evidences also show that cAMP can suppress cell growth by interfering with c-Raf-MEK1/2-Erk signaling pathway [15, 16], attenuating the expression of anti-apoptotic protein Bcl2  or inducing the expression of cell-cycle inhibitor p27kip1 . Moreover, cAMP can stimulate cell differentiation [13, 19] and mesenchymal-to-epithelial transition , which may also lead to cell growth inhibition.
In this study, we show that 8-Br-cAMP at concentration > 1 mM inhibits growth of TNBC but not ER+ cells. Surprisingly, TNBC cell growth was little affected by adenylate cyclase activator forskolin and pan-PDE inhibitor 3-isobutyl-1-methyl-xanthene (IBMX). To elucidate this apparent discrepancy, we uncover that the inability of forskolin/IBMX to inhibit TNBC cell growth is due to a rapid diminution of cellular cAMP by multidrug resistance-associated protein (MRP)-mediated efflux. With the aid of short interfering RNAs (siRNAs), MRP1 and MRP4 are identified as the members of MRP family facilitating rapid cAMP efflux in TNBC cells. Meanwhile, we provide evidences that multiple PDE4 isotypes can diminish cellular cAMP when MRPs are blocked. Finally, we demonstrate that cocktail of forskolin, probenecid (pan-MRP inhibitor) and rolipram (PDE4 inhibitor) effectively inhibit in vitro cell growth and in vivo tumor development of TNBC cells.
A recent study reported that various cAMP-elevating agents were able to inhibit growth of MDA-MB-231, a TNBC line . To determine if the same could be generalized to other breast cancer cell lines, we examined the effect of 8-Br-cAMP, a PDE-resistant cAMP analog, on growth of 4 TNBC and 4 ER+ cell lines. MTT assay showed that 8-Br-cAMP at concentration > 1 mM inhibited growth of TNBC but not ER+ lines (Figure (Figure1A).1A). Further clonogenic assay showed that 1 mM 8-Br-cAMP reduced more than 75% of colonies formed in MDA-MB-231 cells while only 15% reduction the number of in formed colonies was detected in MCF7 cells (Supplementary Figure S1). These results suggest that TNBC cells are selectively sensitive to elevated level of cellular cAMP.
The necessity of 8-Br-cAMP to inhibit TNBC cell growth at concentration > 1mM led us to investigate whether cAMP-elevating agents would be more effective. We treated TNBC cells with forskolin, an adenylyl cyclase activator, and IBMX, a pan-PDE inhibitor alone or together for 4 days followed by cell growth analysis. MTT assay showed that growth of neither TNBC nor ER+ cells was significantly altered by forskolin and IBMX alone or together (Figure (Figure1B1B).
The discrepancy on the effect of TNBC cell growth between high concentration of 8-Br-cAMP and cAMP-elevating agents indicated the possibility that forskolin/IBMX was unable to elevate cellular cAMP to a level sufficient to inhibit TNBC cell growth. To test it, we examined the effect of forskolin on cellular cAMP concentration in both TNBC and ER+ lines. In ER+ cells (BT474, MCF7, T47D and ZR-75-1), forskolin induced a fast and sustained elevation of cellular cAMP in the entire 24-h stimulation period (Figure (Figure2A).2A). In contrast, forskolin transiently increased the level of cellular cAMP which was rapidly diminished after 30 min in TNBC cells (BT549, Hs578T, MDA-MB-231 and MDA-MB-436) and nearly subsided to the level of prior treatment at 1 h (Figure (Figure2B).2B). These results demonstrate that cellular cAMP is quite stable in ER+ cells whereas it is quickly cleared in TNBC cells. To determine the role of PDEs in rapid diminution of cellular cAMP, MDA-MB-231 and MDA-MB-436 cells were pretreated with IBMX prior to forskolin stimulation. Analysis of cAMP in cells showed that IBMX was unable to prevent the clearance of cAMP although the rate was moderately slower (Figure 2C and 2D).
To determine whether efflux is responsible for rapid clearance of cellular cAMP in TNBC cells, we determined what percentage of total cAMP was cellular in MDA-MB-231 and MDA-MB-436 cells at various times of forskolin/IBMX stimulation. Over 90% of total cAMP was detected in cells at 5-min of stimulation and total cAMP in cells reduced to approximately 60–70% at 15 min (Figure (Figure2E).2E). At 1 h, only less than 5% of total cAMP remained in cells (Figure (Figure2E).2E). In parallel, we determined how cAMP was distributed in ER+ MCF7 and T47D cells upon forskolin/IBMX treatment. Similar to TNBC cells, we detected that over 90% of total cAMP was cellular at 5 min but percentage of total cAMP in cells decreased to approximately 80% at 15 min (Figure (Figure2E).2E). Contrary to TNBC cells, there was no further reduction in the percentage of total cAMP that was cellular in both MCF7 and T47D cells (Figure (Figure2E).2E). These results indicate that rapid diminution of cellular cAMP is unique to TNBC cells and is mainly mediated by efflux.
Members of MRP family are efficient cAMP efflux pumps . To investigate the potential role of MRPs in cAMP efflux in TNBC cells, BT549, Hs578t, MDA-MB-231 and MDA-MB-436 cells were pretreated with pan MRP inhibitor probenecid, MPR1-selective inhibitor MK571, MDR inhibitor elacridar hydrochloride, BCRP inhibitor fumitremorgin C or verapamil followed by 1-h of forskolin/IBMX stimulation. More than 70% and approximately 20–35% of total cAMP was found to be cellular in probenecid- and MK571-pretreated cells respectively (Figure (Figure3A).3A). In contrast, none of the other inhibitors increased the percentage of total cAMP over the control (Figure (Figure3A).3A). These results support the notion that MRPs are responsible for cAMP efflux.
To identify the particular MRP family members that facilitate cAMP efflux in TNBC cells, we initially compared the expression of MRP family members in luminal-A, luminal-B, HER2-enriched and basal-like breast tumors using publicly available TCGA dataset. Levels of MRP1, MRP4 and MRP7 mRNA, but not other members of MRP family are significantly higher in basal-like breast tumors compared to those in luminal-A/B breast tumors (Figure (Figure3B3B and Supplementary S2). Since the majority of TNBC are of basal-like phenotype and the majority of tumors expressing ‘basal’ markers are TNBCs , expression of MRP1, MRP4 and MRP7 is most likely greater in TNBCs. Subsequently, we analyzed MRP expression profile on established breast cancer cell lines using datasets available from Cancer Cell Line Encyclopedia. Consistent with results from human breast tumor tissues, MRP1, MRP4 and MRP7 were expressed at least 2-fold higher in TNBC lines than ER+ ones (Figure (Figure3C).3C). To validate the findings derived from public datasets, we examined the expression of MRP family members in a panel of breast cancer cell lines. QRT-PCR showed that MRP2, MRP6, MRP8 and MRP9 were hardly detectable in all lines while MRP3 and MRP7 were similarly expressed in ER+ and TNBC lines and level of MRP5 mRNA was higher in ER+ lines (Supplementary Figure S3). In contrast, expression of MRP1 and MRP4 was much greater in TNBC lines than ER+ ones (Figure (Figure3D).3D). Similarly, western blot also showed that levels of MRP1 and MRP4 were higher in TNBC lines than ER+ ones (Figure (Figure3E3E).
We next investigated the effect of depleting MRP1, MRP4 or MRP7 on cAMP efflux by introducing their respective siRNA pools into MDA-MB-231 and MDA-MB-436 cells (Figure (Figure4A).4A). Knockdown of MRP1 rendered 21 and 16% of total cAMP cellular in MDA-MB-231 and MDA-MB-436 cells respectively while knockdown of MRP4 kept 52 and 60% of total cAMP in cells respectively (Figure (Figure4B).4B). Contrarily, knockdown of MRP7 displayed little effect on the level of cellular cAMP (Figure (Figure4B).4B). When both MRP1 and MRP4 were silenced together, over 70% of total cAMP wwas found to be cellular in these two lines (Figure (Figure4B).4B). These results suggest that MRP4 is the principal cAMP efflux pump in TNBC cells whereas MRP1 also contributes to this event.
The observation that IBMX reduced the rate of cAMP clearance in TNBC cells (Figure 2C and 2D) implicated a possible role of PDEs in cAMP diminution. To test it, BT549, Hs578T, MDA-MB-231, MDA-MB-436 cells were pretreated with probenecid and then stimulated with forskolin in the absence or presence of IBMX for 1 h followed by analyzing amount of cellular cAMP. Level of cellular cAMP in cells co-treated with forskolin and IBMX was nearly four times higher than that in cells treated with only forskolin (Figure (Figure5A).5A). These results support the notion that PDEs are involved in cAMP clearance in TNBC cells.
There are 11 families of PDEs and 8 of them can degrade cAMP . To identify the particular PDEs mediating cAMP clearance in forskolin-stimulated TNBC cells, probenecid-pretreated MDA-MB-231 cells were stimulated with forskolin in the presence of inhibitors specific for PDE1, 2, 3, 4, 7, 8 or 10 for 1 h. Compared to cells stimulated with only forskolin, level of cellular cAMP was almost tripled in cells co-treated with forskolin and rolipram (PDE4 inhibitor) but not significantly altered by forskolin with any other inhibitors (Figure (Figure5B).5B). Identical results were also obtained with MDA-MB-436 cells (Supplementary Figure S4). These results suggest that PDE4 family members decompose cellular cAMP in forskolin-stimulated TNBC cells.
As there are four isotypes (PDE4A, 4B, 4C and 4D) in PDE4 family , we analyzed the expression of PDE4 isotypes in various breast tumor types with TCGA dataset. The expression of PDE4B was significantly higher in basal-like breast tumors compared with other breast tumor types (Figure (Figure5C).5C). Surprisingly, further analysis of established breast cancer cell lines using datasets from Cancer Cell Line Encyclopedia showed that PDE4A, rather than PDE4B was overexpressed in TNBC cell lines compared with ER+ ones (Figure (Figure5D).5D). We also measured the levels of PDE4 isotypes in both TNBC and ER+ breast cancer cell lines by qRT-PCR. While PDE4C was not detectable in any of these lines, we did not observe a clear difference in PDE4B expression between TNBC and ER+ cells (Figure (Figure5E).5E). However, both PDE4A and PDE4D were expressed in a higher level in TNBC lines compared with ER+ lines (Figure (Figure5E).5E). To experimentally verify PDE4 isotypes involved in cAMP diminution, we depleted PDE4A, PDE4B or PDE4D in MDA-MB-231 cells with the aid of respective siRNA pools. Upon the co-treatment of forskolin and probenecid, we observed that silencing the expression of any of them enhanced the level of cellular cAMP in some extent (Figure (Figure5F).5F). Since the abundance of cellular cAMP in cells with depletion of three PDE4 isotypes reached to that observed in rolipram-treated cells (compare Figure Figure5F5F with 5B), these results suggest that multiple PDE4 isotypes are involved in the clearance of cellular cAMP in TNBC cells.
The ability of MRP and PDE4 inhibitors to prevent rapid cellular cAMP diminution in forskolin-stimulated TNBC cells (Figure (Figure5B)5B) provided a window of possibility to suppress TNBC cell growth by combined treatment of forskolin, probenecid and rolipram. To test this possibility, MDA-MB-231 and MDA-MB-436 cells were treated with these agents alone or together for 4 days followed by MTT assay to monitor cell growth. Forskolin displayed little effect on cell growth while probenecid and rolipram were able to slightly inhibit cell growth (Figure (Figure6A).6A). In contrast, treating cells with them together led to over 80% of reduction in cell growth (Figure (Figure6A).6A). These results are consistent with notion that sustaining elevated level of cellular cAMP can lead to the suppression of TNBC cell growth.
To elucidate the molecular mechanism underlying cAMP-induced growth inhibition, MDA-MB-231 and MDA-MB-436 cells were co-treated with forskolin, probenecid and rolipram or left untreated for 1 day followed the analysis of cell cycle progression. Flow cytometry showed that treatment increased population of cells at G0/G1 from approximately 50% to 63% in MDA-MB-231 and 48% to 62% in MDA-MB-436 cells (Figure (Figure6B),6B), indicating a cell cycle arrest at G1 phase. Cell cycle arrest at G1 phase was also consistent with the observation that, similar to 5 mM 8-Br-cAMP, this treatment reduced the abundance of cyclin D1, cyclin E2 and PCNA whereas increased the amount of p21 and p27 (Figure (Figure6C).6C). Since the same treatment did not increase the number of apoptotic cells judging by Annexin/PI-based flow cytometry and PARP cleavage (Supplementary Figure S5), these results suggest that cAMP suppresses TNBC cell growth by arresting cell cycle progression.
The effectiveness of combined forskolin, probenecid and rolipram treatment to suppress TNBC cell growth led us to investigate the potential of this combination treatment to deter TNBC tumor development with the aid of the well-established orthotopic breast tumor model [25–27]. MDA-MB-231 and MDA-MB-436 cells were injected into mammary fat pad area of female nude mice for 1 week, mice were then randomly divided into 5 groups and each received vehicle, forskolin, probenecid, rolipram or cocktail of all three compounds 3 times a week for 6 weeks (Figure (Figure7A).7A). The dose of forskolin chosen for animal study has previously shown to effectively elevate cellular cAMP level in experimental mouse myeloma model  while the doses of probenecid and rolipram was based on their ability to block their respective targets in in vivo study . Tumors were evident in all mice at the onset of therapy and progresssed rapidly in mice receiving vehicle (Figure (Figure7B).7B). Administering mice with forskolin, probenecid or rolipram slightly slowed down tumor development (Figure (Figure7B).7B). Strikingly, tumor development almost completely ceased 3 weeks after receiving the cocktail (Figure (Figure7B).7B). At the end of treatment, we weighed tumors collected from sacrificed mice. Compared to control mice, we observed approximately 80% reduction in tumor weight in mice that received cocktail (Figure (Figure7C).7C). In a parallel experiment, we performed identical experiment with ER+ MCF7 cells. While tumor outgrowth was slower with MCF7 cells in comparison with MDA-MB-231 or MDA-MB-436 cells, we only detected moderately suppressive effect in tumor development or final tumor weight in mice receiving cocktail (Supplementary Figure S6A and S6B). These results suggest that cocktail of forskolin, probenecid and rolipram specifically suppress TNBC tumor development.
To link ceased tumor outgrowth to arrested cell cycle progression, we performed immunohistochemistry to examine the intensity of cyclin D1 and p21 staining on collected tumors. Strong cyclin D1 staining was detected in tumors derived from control mice whereas staining of p21 was very weak (Figure (Figure7D).7D). In contrast, tumors derived from mice administered with cocktail displayed little cyclin D1 but robust p21 staining (Figure (Figure7D).7D). These results further support the notion that the stoppage of tumor outgrowth in mice receiving cocktail is the consequence of TNBC cell cycle arrest.
Breast cancer can be classified into four main types on the basis of the presence/absence of several receptors: ER+, PR+, HER2-enriched and TNBC. Hormone therapy is effective for the treatment of most ER+/PR+ breast cancer while HER2-targeted therapy is successfully applied to HER2-enriched breast cancer [30, 31]. However, TNBC-targeted therapy is not available and cytotoxic chemotherapy is the only approved therapy for TNBC. Unfortunately, TNBC patients often develop drug resistance and demise with subsequent metastasis . Since cAMP-elevating agents have been shown to effectively suppress growth of various cancer cell types [7–9], we explored the potential of targeting TNBC with such agents. In this study, we showed that high concentration of cAMP analog 8-Br-cAMP (> 1 mM) specifically suppressed TNBC cell growth (Figure (Figure1A).1A). This finding is in agreement with an early report in which 8-Br-cAMP was shown to inhibit growth of TNBC MDA-MB-231 cells at a concentration above 1mM . Since it is infeasible to use almost any agent at such high dose in clinic, we tested the potential of adenylate cyclase activator forskolin and PDE inhibitor IBMX as an alternative. Unfortunately, neither alone or together exhibited significant inhibitory effect on TNBC cell growth (Figure (Figure1B1B).
Adenylate cyclase-mediated production and PDE-mediated degradation are thought to be the mechanism controlling the level of cellular cAMP. We revealed that forskolin quickly induced production of cAMP in both TNBC and ER+ breast cancer cells (Figure (Figure2A).2A). However, a rapid diminution of cellular cAMP followed the initial cAMP production in TNBC but not ER+ cells. The rapid clearance of cellular cAMP was obviously not due to PDE-mediated decomposition because pan-PDE inhibitor IBMX only slightly slowed the rate of cellular cAMP diminution (Figure (Figure2).2). Instead, we revealed that MRP1/4-mediated efflux was a principal mechanism eradicating cellular cAMP while PDE4A was also able to decompose cAMP when MRPs were blocked (Figures (Figures3,3, ,44 and and5).5). Our results thus add MRP-mediated efflux as a principal mechanism TNBC cells employ to control cellular cAMP level.
Early studies have reported that agents raising cAMP are able to suppress growth of colon and medullary thyroid cancer cells [32, 33]. However, we found that commonly-used cAMP-elevating agents exhibited little effect on TNBC cell growth (Figure (Figure1).1). We reason that inability of cAMP-elevating agent to deter TNBC cell growth was due to the coordinated action of rapid MRP1/4-mediated cAMP efflux and PDE4-mediated cAMP decomposition. This is apparently the case because forskolin is only able to significantly curb TNBC cell growth in the presence of MRP and PDE4 inhibitors (Figure (Figure6).6). Similarly, we found that only the cocktail of forskolin, probenecid and rolipram was able to cease TNBC tumor outgrowth (Figure (Figure7).7). Early studies indicate that cAMP might inhibit cell growth by inducing cell cycle arrest or apoptosis [7, 9, 34]. Our results pinpoint the former as the mechanism because combined treatment of forskolin, probenecid and rolipram arrested cell cycle at G1 phase but did not induce apoptosis (Figure (Figure66 and Supplementary Figure S5). Moreover, this treatment also decreased levels of cyclin D1/E2 while increased abundance of p21 and p27 in TNBC cells (Figure (Figure66).
Our results showed that only growth of TNBC but not ER+ breast cancer cells was sensitive to elevated level of cellular cAMP (Figures (Figures1,1, ,66 and and7).7). TNBC cells’ selective sensitivity to elevated level of cAMP may explain why cAMP is rapidly diminished in these cells but not in ER+ cells (Figure (Figure2).2). As knockdown of MRP1/4 largely prevented rapid efflux of cellular cAMP (Figure (Figure4)4) while depletion of PDE4A/B/D blocked cAMP decomposition, we conclude that MRP1/4 and multiple PDE4 isotypes work in concert to diminish cAMP in TNBC cells. MRP1 and 4 are overexpressed in basal-like breast tumors and established TNBC cell lines (Figure (Figure3).3). This observation is consistent with their role to facilitate cAMP diminution in TNBC cells. Our findings suggested the possibility of targeting TNBC by raising the level of cellular cAMP through interfering with both MRP1/4 and PDE4 functions. This possibility is evidently supported by our observation that cocktail of forskolin, probenecid and rolipram at clinically reasonable doses suppressed TNBC, but not ER+ breast tumor outgrowth (Figure (Figure7).7). As inhibitors specific for MRP and PDE4 are approved for clinic use, our study has thus laid a foundation for a novel TNBC-targeted therapy with these compounds.
All cell lines were purchased from American Tissue Culture Collection within the past 6 months and cultured in DMEM supplemented with 10% fetal bovine serum. Information on all agonists and inhibitors is provided in Supplemental Materials. Control siRNA, MRP and PDE4 siRNA pools were purchased from GE Dharmacon (Lafayette, CO).
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was performed to analyze cell growth as previously described [35–37]. In each assay, 5x103 cells were seeded into each well of 24-well culture plates for overnight followed by treatment of 8-Br-cAMP or other agents for 4 days. Clonogenic assay was performed to determine the inhibitory effect of high concentration of 8-Br-cAMP. Briefly, 5 × 103 cells were seeded in 6-cm dishes in the absence or presence of 1 mM 8-Br-cAMP for 7 days and media were replaced once every other day. Cells were fixed with 0.5% glutaraldehyde and then stained with 0.05% crystal violet. Colonies were counted under a dissecting scope.
Amount of cAMP was analyzed using Cyclic AMP Assay Kit (Cell Signaling Technology, Danvers, MA). Briefly, overnight-cultured cells were pretreated with PDE inhibitors, probenecid or other inhibitors for 1 h and then stimulated with forskolin for varying length of times. Both cells and media were collected for cAMP measurement. cAMPs in cells and media were considered as cellular and secreted cAMP respectively. To investigate the importance of MRPs or PDE4s, cells were transfected with control siRNA or siRNA pools for distinct MRP or PDE isotype for 4 days prior to the analysis of cAMP.
To analyze the status of cell cycle, cells were detached with trypsin, washed and fixed in 100% ethanol. After a brief centrifugation to remove ethanol, cells were suspended in PBS containing 20 μg/mL of propidium iodide followed by flow cytometry analysis using FACSCanto II flow cytometer (BD Biosciences, Bedford, MA). The data were analyzed using the BD FACSDiva Software.
Western blotting was performed as previously described . Information on all antibodies is provided in Supplemental Materials. Level of β-actin was determined for every blot as an internal loading standard.
All procedures were conducted with animal welfare considerations and approved by the Ethical Committee of Shanghai University of Traditional Chinese Medicine and performed as previously described [25–27]. Briefly, 1 × 106 cells were injected into 4th mammary fat pad of 6-week-old female BALB/c nude mice (Academia Sinica, Shanghai, China). After 1 week, mice were randomly divided into five groups (6 mice/group) and received treatment 3 times a week for 6 weeks. Five groups were vehicle (control), forskolin, probenecid, rolipram, and cocktail of forskolin, probenecid and rolipram. Tumor development was monitored by weekly measuring tumor volumes (V) which were calculated using formula of V = 0.5 x Dmax x (Dmin)2, where Dmax is the maximal tumor diameter and Dmin is the corresponding perpendicular diameter.
Statistical differences were calculated using 2-tailed Student t test. The expression of MRP family members and PDE4 isotypes in various breast tumor types was evaluated using dataset TCGA_BRCA_exp_HiSeqV2 (https://genome-cancer.ucsc.edu) and shown in box-and-whisker plots. Interquartile range (IQR) was expressed by the colored box and the bar indicated the median value. Statistical difference was calculated using Analysis of variance. To compare the expression of MRP family members and PDE4 isotypes between TNBC and ER+ breast cancer cell lines, expression data of various breast cancer cell lines from Cancer Cell Line Encyclopedia (http://www.broadinstitute.org/ccle/home) were retrieved. CCLE_Expression_2012-10-18.res and CCLE_Expressoin_2012-09-29.res datasets were respectively used to evaluate the expression of MRP family members and PDE4 isotypes. Heat maps (green-black-red, representing low-medium-high expression respectively) were constructed by Gene Cluster 3.0.
This study was supported by grants from E-Institutes of Shanghai Municipal Education Commission (Project E03008), “085” First-Class Discipline Construction Innovation Science and Technology Support Project of Shanghai University of TCM (085ZY1206), National Science Foundation of China (31229002) and NIH CA187152.
CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.