In the postgenomic era of cancer research, there is an intensive discussion to what extent the development of metastases is genetically based, and what role can be ascribed to environmental factors. On one hand, Bernards and Weinberg propose that mutant genes confer a Darwinian selective advantage for a clonal selection of metastatic tumor cells.1 Their theory is supported by the finding that certain mutations, e.g. mutations of the tumor-suppressor gene MADH4, occur more frequently in metastatic tumors.2 On the other hand, the comparative gene expression profiling of primary breast tumors and distant metastases showed striking similarity, suggesting that the metastatic capability in breast cancer is an inherent feature and is not based on clonal selection.3 Recent research has shown that ligands to serpentine receptors play an important role in the regulation of tumor cell migration. These ligands predominantly consist of 2 groups, the chemokines and the neurotransmitters.4 Among the chemokines, the stromal cell-derived factor-1 (SDF-1) is best investigated for its role in metastasis development,5 and in the localization of metastases as shown in mice.6 Within the group of neurotransmitters, norepinephrine is one of the most potent known stimulators for the migration of tumor cells. We have shown previously in this journal and elsewhere that norepinephrine induces migration in breast, colon and prostate carcinoma cells, and that this induction can be inhibited by β2-adrenoceptor blocking drugs in vitro.7, 8, 9
In this present study, we address the question whether the previously observed increase of tumor cell migration in response to norepinephrine in vitro corresponds to an in vivo increase of metastasis development in mice, and whether this effect can be inhibited by the chemopreventive treatment with clinically established β-blockers.
Material and methods
Cell culture and proliferation assay
PC-3 human prostate carcinoma cells (DSMZ, Braunschweig, Germany) were cultured in HAMs and RPMI medium (1:1) containing 10% heat-inactivated fetal calf serum (all medium components were derived from PAA, Linz, Austria) at 37°C and 5% CO2 humidified atmosphere, as described previously.9 The cells were transfected with a green fluorescence protein (GFP)-coupled luciferase expression vector (pGFPLuc, BD Biosciences, Palo Alto, CA), using Ex-gene 500 (Fermentas, Sankt Leon-Rot, Germany). This vector contains a neomycin resistance site. Accordingly, a stable line of transfected cells was grown in selection medium, culture medium with 400 μg/ml G418 (PAA), and sorted for high GFP expression using a FacsCalibur flow cytometer (Becton Dickinson, Heidelberg, Germany). All experiments presented herein were conducted with these pGFPLuc-transfected PC-3 cells (designated as PC-3-luc).
For the investigation of proliferation, aliquots of the PC-3-luc cells were cultured under the conditions, as described earlier. Norepinephrine (arterenol, 10 μM; Sigma-Aldrich, Deisenhofen, Germany), propranolol (10 μM; Calbiochem, Bad Soden, Germany) or a combination of both were added daily to the culture medium. After 2 days of incubation, the medium was renewed. After another 2 days, at the subconfluent state of growth, aliquots of the cells in each sample were counted.
Cell migration experiments
The three-dimensional cell migration assay was performed, as described in detail previously.7, 8, 9 In brief, 5 × 104 PC-3-luc cells were mixed with 150 μl of a buffered collagen solution (1.63 mg/ml collagen type I; Collagen, Freemont, CA) containing minimal essential medium (Sigma-Aldrich), as well as the investigated substances norepinephrine and propranolol. This collagen solution was filled into self-constructed chambers. After polymerization of the collagen, the migratory activity was recorded by time-lapse videomicroscopy for 12 hr.
For the analysis of the migratory activity, 30 cells of each recording were randomly selected and the migration paths were digitized in 15 minute intervals by computer-assisted cell tracking. The locomotory active part of the population was calculated for each time point.
Experiments with BALB/c nude mice
Four- to 5-week-old athymic female BALB/c nude mice were provided by Charles River Wiga (Sulzfeld, Germany). We used only female mice, as male mice are supposed to be under stress conditions from the caging with mates. The mice were kept under pathogen-free conditions in type IV Makrolon cages (5 mice per cage) in an air flow cabinet (Uni-Protect, Ehret, Emmendingen, Germany) at 23°C, 12 h/12 h day/night cycle; water and food ad libidum. The heart rate of awake and narcotized mice was measured before microosmotic pump changes, using a Sirecust 341 electrocardiogramm device (Siemens, Munich, Germany).
PC-3-luc cells were injected into the right thigh at the age of 6 weeks; 1 × 106 cells were injected per mouse. Five mice were used per experiment and sample. After 1 week, microosmotic pumps (Alzet model 1002, Durect, Cupertino, CA) were implanted subcutaneously on the back of the mice. We used these pumps for the application of the substances, because norepinephrine especially is quickly metabolized and thus has a short half-life. Therefore, in contrast to other application methods (e.g. repeated injection), a continuous source of substances is applied by the microosmotic pumps, resulting in a constant level over time. The pumps were filled with physiological saline containing 28 mM norepinephrine, 28 mM propranolol or a combination of both (which equals a daily dose of 1 μmol/100 g bodyweight10), as well as ascorbic acid (0.2%; Sigma-Aldrich) as a preservative. The pumps of the control group were filled with the saline solution and ascorbic acid. Implantation of the pumps was conducted under Ketanest S (Pfizer, Karlsruhe, Germany) and Rompun (Bayer, Leverkusen, Germany) narcosis. After 2 weeks, the pumps were renewed. Before the pumps were changed, luciferin was intraperitoneally injected and the tumor cells were detected by in vivo imaging,11 utilizing a Hamamatsu C4742-98 system (Hamamatsu, Herrsching, Germany). After another 2 weeks, the mice were narcotized again, in vivo imaging was performed, and the mice were sacrificed by cervical dislocation under narcosis.
The animal experiments with the athymic BALB/c nude mice were accredited by the district government Arnsberg (Germany), approved by the local ethics animal protection committee and are conform to the relevant regulatory standards.
Human DNA within the mouse organs was detected by quantitative rtPCR.12 The detection is based on a human-specific α-satellite DNA sequence. For the preparation of the samples, the organs were removed from the sacrificied mice, homogenized, and the DNA was extracted using a DNeasy tissue kit (Qiagen, Hilden, Germany). Quantitative rtPCR was performed utilizing an ABI PRISM 7700 Sequence Detector (PE Biosystems, Foster City, CA), according to the protocol established by Becker and coworkers.12
Standard curves were made by directly injecting a known number of PC-3-luc cells into organs (lung, liver and spleen). The organs were derived from untreated mice. The standard curve of the spleen was used to calculate the absolute number of PC-3-luc cells present in the lumbar lymph node metastases. The detection limit of the rtPCR is less than 10 PC-3-luc cells in 10 mg of mouse lung tissue and less than 100 PC-3-luc cells in 10 mg of mouse liver and spleen tissue.
Slides with human tissue samples were provided by BioCat (Heidelberg, Germany). The microarrays contained frozen sections of various normal tissues samples and tumor samples of the respective tissues. The sections were treated with 0.3% H2O2 for 30 min at room temperature. After washing with PBS, the sections were blocked with 10% normal bovine serum (PAA) for 1 hr at room temperature, and subsequently with a rabbit anti-human β2-adrenoceptor antibody (Santa Cruz Biotechnology, Santa Cruz, CA) at 4°C overnight. After a further washing step with PBS, the sections were incubated with a Cy3-coupled secondary goat anti-rabbit antibody (Dianova, Hamburg, Germany) for 1 hr at room temperature. After intensive washing with PBS, the fluorescence of the samples was analyzed using a dual-laser microarray scanner (GenePix 4000, Axon Instruments, Union City, CA) at 532 nm.
Statistical significance was calculated using the Student's t test (two-tailed and unpaired).
Growth and migration of PC-3-luc cells in response to norepinephrine and propranolol in vitro
Incubation of PC-3-luc human prostate carcinoma cells with either 10 μM norepinephrine, 10 μM propranolol or a combination of both had no measurable influence on the proliferation of the cells (control, 3,978 ± 839 cells; norepinephrine, 4,142 ± 322 cells; propranolol, 3,758 ± 72 cells and norepinephrine and propranolol, 3,804 ± 326 cells; Fig. 1a). In contrast, treatment of the cells with 10 μM norepinephrine led to a significant (p = 0.008) increase of the migratory activity from 15.7 ± 3.0% to 29.0 ± 3.6% locomoting cells (Fig. 1b). This increase was significantly inhibited by the additional treatment of the cells with an equimolar amount of propranolol (17.1% ± 2.2% locomoting cells; p = 0.008). Propranolol alone had no effect on cell migration (16.2% ± 4.5% locomoting cells).
Weight and heart rate of mice treated with norepinephrine and propranolol
After injection of PC-3-luc cells to 6-week-old athymic BALB/c nude mice, the animals showed a significant loss of weight in comparison to mice without tumor cell injection from the age of 9 weeks on (p < 0.001; Fig. 2a). However, neither norepinephrine nor propranolol applied by microosmotic pumps had an effect on the weight of the animals (Fig. 2b). As these substances are in clinical use for their regulatory influence on the cardiovascular system, we measured the heart rate of the mice to monitor their pharmaceutical effectiveness. We observed changes in the heart rate by both substances, however, with differences depending on whether the mice were narcotized or not (Fig. 2c). Norepinephrine did not increase the heart rate in awake mice (710 ± 22 beats/min vs. control 719 ± 9 beats/min; Fig. 2c, left), but propranolol alone significantly (p = 0.002) decreased the heart rate to 626 ± 52 beats/min and abolished the effect of norepinephrine (632 ± 60 beats/min; p = 0.006). No effect of norepinephrine in comparison to the control was observed on the heart rate in awake mice; this might be due to the agitation of the mice from the procedure of the electrocardiographic measurement.
In contrast, we observed a significant augmenting effect of norepinephrine on the heart rate in narcotized animals (from 159 ± 25 to 197 ± 34 beats/min; p < 0.001; Fig. 2c, right). Propranolol likewise inhibited this effect (167 ± 15 beats/min). However, as the narcotic drugs had a strong decreasing effect on the heart rate, we observed no further decrease of the heart rate under control levels by the treatment with propranolol alone (172 ± 33 beats/min).
Tumor growth and metastasis development of PC-3-luc cells in BALB/c nude mice
PC-3-luc cells were intramuscularly injected into the right thighs of the mice and the growth of the tumors was analyzed by in vivo imaging using luciferase/luciferin (Fig. 3a). In accordance with the in vitro findings, the growth of the primary tumor of PC-3-luc cells in the thighs of the mice was not influenced by norepinephrine or propranolol (Fig. 3b). However, the development of lumbar lymph node metastases changed under the influence of both norepinephrine and propranolol (Fig. 3c). Five weeks after injection of the PC-3-luc cells, the mice treated with norepinephrine had 165% ± 69% larger lymph node metastases in the abdomen than the control group, whereas the treatment with propranolol inhibited this effect down to 77% ± 6% of the control. Both of these effects were statistically significant (p = 0.014 and p = 0.009, respectively); propranolol alone had no significant effect (84% ± 20% of the control). After 3 weeks, lumbar (periaortic) lymph node metastases were clearly detectable in some individual mice within the group treated with norepinephrine, while in the mice of all other groups only very weak fluorescence was detectable in the lumbar lymph nodes (Fig. 3c). This shows that in norepinephrine-treated mice lumbar lymph node metastases occur much earlier than in the other groups, which we conclude is due to the higher migratory activity also observed in vitro.
After sacrificing the animals, the brain, lung, liver, adrenal gland and lumbar as well as caudal lymph nodes (Fig. 4a) were removed and subjected to quantitative rtPCR. Using this quantifying rtPCR, we analyzed the human DNA within the mice tissues by an α-satellite sequence specific for humans. Of all the investigated organs, the lumbar lymph nodes were the only tissue with detectable amounts of human DNA. In control experiments, we directly injected various amounts of PC-3-luc cells into 10 mg of spleen tissue and subjected these tissues to quantitative rtPCR likewise. The resulting correlation between the number of injected cells to the rtPCR value was used as a standard curve to calculate the number of PC-3-luc cells in the lumbar lymph nodes of the mice (Fig. 4b). The quantitative rtPCR showed that the number of PC-3-luc cells in the lymph nodes of the control group was 14.2 ± 0.8 million cells per animal, and in the mice treated with norepinephrine 25.2 ± 5.6 million cells per animal, which is a significant increase of 77% (p = 0.027; Fig. 4c). Propranolol alone significantly reduced the number of PC-3-luc cells found in the lymph nodes to 4.3 ± 2.6 million cells per animal (p = 0.003), and significantly inhibited the norepinephrine effect (8.4 ± 3.9 million cells per animal, p = 0.013; Fig. 4c). As the number of PC-3-luc cells that we have measured by rtPCR was similar among the 3 experiments (Fig. 4c), the nearly 3-fold higher luminescence of the in vivo imaging in experiment 2 (white bars, Figs. 3b and 3c) probably results from a higher luciferin expression of the PC-3-luc cells used in this experiment.
Expression of the β2-adrenoceptor in human tissue
We have shown previously using breast, colon and prostate tumor cell lines that the promigratory effect of norepinephrine can be inhibited by a specific blockade of the β2-adrenoceptor.7, 8, 9 As a complement to the aforementioned results regarding the in vivo relevance of the promigratory effect of norepinephrine in mice, we investigated the expression of the β2-adrenoceptor in normal human breast, colon and prostate tissue and in 10 tumors of the respective tissues (Fig. 5). Lung tissue served as a positive control, as lung epithelium is known to express the β2-adrenoceptor.13 The commercially available human tissue arrays showed that the β2-adrenoceptor expression is restricted to certain cells in normal breast tissue. In normal prostate tissue, no expression of this receptor was detectable, and in normal colon epithelium, the β2-adrenoceptor was highly expressed. Interestingly, the β2-adrenoceptor was expressed in 8 of 10 samples of the breast carcinoma tissue, in 7 of 10 samples of the prostate carcinoma tissues and in 9 of 10 samples of the colon carcinoma tissue (examples are shown in Fig. 5, lower lane). These data provide further support for the potential clinical impact of our results on the function of norepinephrine in metastasis development.
Psychosocial stress has been ascribed a role in the incidence and progression of cancer since 1926.14 The molecular basis of this interconnection between malignant tumors and the neuroendocrine system was for a long time unclear. The immune system has been attributed a function as a mediator between these two systems, and as a consequence of this mediating role, immunosuppression was supposed to enhance the establishment of a tumor.15, 16 Accordingly, we have shown previously that the cytotoxicity of natural killer cells is strongly impaired by norepinephrine,17 which might be supportive for the effects on the metastasis formation that we have described herein; we have not observed further effects on the function of natural killer cells or cytotoxic T lymphocytes in vitro.17 In general, an immunosuppression by norepinephrine or other neurotransmitters might add effects to the metastasis formation. Of course, we can not separate these effects in a complex living system, although we have used athymic mice that lack important parts of the cellular immune system. Thereby, we reduced the influence of the immune system. However, current research has revealed that several neurotransmitters have direct influence on the migratory activity and invasiveness of tumor cells,18 and thus do not need the immune system as a mediator. We have shown previously in vitro that stress-related neurotransmitters are the most potent direct stimulators for the migration of carcinoma cells of various tissue origins. These neurotransmitters are norepinephrine,7, 8, 9 dopamine9 and substance P.4 Furthermore, with regard to norepinephrine, we have shown by flow-cytometry that both the β1- and β2-adrenoceptors are expressed on the surface of colon,7 breast8 and prostate9 carcinoma cells. The use of well characterized and subtype specific pharmacological inhibitors (e.g. atenolol for the blockade of the β1-adrenoceptor) revealed that the promigratory effect was in either case mediated by the β2-adrenoceptor.7, 8, 9 Of all carcinoma cell lines that we have investigated in vitro for their susceptibility to norepinephrine, the PC-3 prostate carcinoma cells were those that were best characterized in in vivo mouse models.19 Furthermore, PC-3 cells had the highest proliferation rate in vitro, and when we started the mouse experiments, the PC-3 cells turned out to be the fastest growing cells in vivo, too. This was an important argument to select this cell line with regard to experimental limitations, as the microosmotic pumps were drained and had to be renewed after 2 weeks. We decided to change the pumps only once so as to minimize potential side-effects on the experimental system due to surgery.
The results presented herein deliver evidence for a migration-based increase of metastasis development by norepinephrine in mice, in accordance with the aforementioned experimental in vitro data. We show that the growth of the tumor cells is not influenced by norepinephrine in vitro or in vivo, as observed by the similar growth of control cells and norepinephrine-treated cells in both cell cultures (Fig. 1a) and primary tumors (Fig. 3b). In contrast, the migratory activity of the tumor cells is increased under the treatment with norepinephrine. This increase in migratory activity leads to an augmented and earlier emigration of tumor cells from the primary tumor and into the lumbar lymph nodes, and in consequence to larger metastases, as we have seen by in vivo imaging and rtPCR 5 weeks after tumor injection. Whether norepinephrine leads to organ-specific metastasis to other norepinephrine-rich organs must be shown in further experiments with an enhanced study design. This observation that tumor cells migrate to the lymph nodes (in this case from the primary tumor in the thigh) is in line with the frequent clinical observation of metastasis development. The lymph nodes and other organs of the immune system, such as bone marrow, thymus and spleen,20 contain noradrenergic and neuropeptidergic nerve fibers. Whether norepinephrine or other neurotransmitters are responsible for the directed metastatic migration towards the lymph nodes also remains to be investigated. We have used this model of an intramuscular injection of tumor cells for the investigation of metastasis development, because the mice initially develop solid primary tumors. Other mice models are termed as metastasis models, too, e.g. the intravenous injection. However, in these models, tumor cells are unspecifically enriched in organs, which are highly supplied with blood and are thus largely based on proliferation rather than on migration.21 Furthermore, it has been described recently that PC-3 cells are tumorigenic and metastatogenic in mouse, even when not applied orthotopically.19
The promigratory effect of norepinephrine in vivo was inhibited by the well-characterized β-blocker propranolol, which has been in safe and effective clinical use for decades for the treatment of hypertension. Application of propranolol abolished the prometastatic effect of the norepinephrine, which was released by microosmotic pumps. Furthermore, propranolol reduced the metastases formation to less than control levels, which is likely due to the blocking of endogenously produced norepinephrine by the mice. These results suggest the clinical testing for the use of β-blockers for the chemoprevention of metastasis development in patients with diagnosed cancer, especially with regard to the fact that the diagnosis of cancer itself and the according clinical treatment causes stress. We have shown previously that the stimulating effect by norepinephrine on tumor cells migration in vitro is mediated by β2-adrenoceptors.7, 8, 9 Accordingly, we have used the unspecific β1/2-blocker propranolol instead of β1-specific blockers in the in vivo experiments we present here. Furthermore, we have shown by immunohistochemistry that the β2-adrenoceptor is expressed in human prostate carcinoma tissue. In addition to prostate tissue, we present in Figure 5 the immunostainings of samples from breast and colon tissue, as these are the two other types of carcinomas, which we have identified to respond to norepinephrine with increased migratory activity.7, 8 The combination of our in vivo and in vitro results delivers good prospects for the development of nonheart active, β2-specific blockers for the inhibition of metastasis development.
It is noteworthy, though no epidemiological study has been designed so far to analyze the role of β-blockers in the progression of cancer, that there are results that a treatment with β-blockers can reduce the incidence of cancer: Algazi et al. have shown a reduced cancer risk for patients taking β-blockers in a French study.22 Likewise, Perron et al. have shown the same correlation focusing on prostate cancer in a Canadian study.23 In both the studies, the authors found a positive correlation between the duration of the intake of β-blockers and the extent of the reduction of cancer risk. Unfortunately, no discrimination between unspecific β1/2-adrenoceptor blockers and β1-specific blockers had been made in these studies. However, these epidemiological results provide evidence that norepinephrine and β-blockers not only play a role in tumor progression and secondary prevention––as we have shown herein––but also in tumor incidence and primary prevention.
In summary, our results further support the view that the development of metastases is not merely genetically determined, but is under the strict influence of the organism's own signal substances. As a practical conclusion with potential impact on the treatment of cancer, these results show the need for more research on the role of neurotransmitters in specific types of cancer, and provide further justification for the investigation of the use of β-blockers for the chemopreventive inhibition of metastasis development.
We thank Beate Mainusch for excellent technical assistance and mouse care.