The effect of thrombin on tumor cell cycle activation and spontaneous growth was examined in synchronized serum-starved tumor cell lines and a model of spontaneous prostate cancer development in TRAMP mice. BrdUrd incorporation and propidium iodide staining of prostate LNCaP cells arrested in G0 and treated with thrombin or serum revealed a 48- and 29-fold increase in S phase cells, respectively, at 8 hours. Similar results were obtained with TRAMP cells and a glioblastoma cell line, T98G. Cell cycle kinases and inhibitors in synchronized tumor cells revealed high levels of p27Kip1 and low levels of Skp2 and cyclins D1 and A. Addition of thrombin, TFLLRN, or serum down-regulated p27Kip1 with concomitant induction of Skp2, Cyclin D1, and Cyclin A with similar kinetics. LNCaP p27Kip1-transfected cells or Skp2 knockdown cells were refractory to thrombin-induced cell cycle activation. MicroRNA 222, an inhibitor of p27Kip1, was robustly up-regulated by thrombin. The in vitro observations were tested in vivo with transgenic TRAMP mice. Repetitive thrombin injection enhanced prostate tumor volume 6- to 8-fold (P < 0.04). Repetitive hirudin, a specific potent antithrombin, decreased tumor volume 13- to 24-fold (P < 0.04). Thus, thrombin stimulates tumor cell growth in vivo by down-regulation of p27Kip1. [Cancer Res 2009;69(8):3374–81]
- cell cycle
Experimental data from numerous reports ( 1– 19) suggest that thrombin contributes to a more malignant phenotype in vivo by activating tumor-platelet adhesion, tumor adhesion to subendothelial matrix, tumor implantation, tumor growth, experimental pulmonary metastasis, and tumor-associated angiogenesis. However, there is no direct proof that thrombin enhances primary tumor growth in vivo because this is dependent on tumor implantation, angiogenesis, and metastasis. Neither is there any in vitro data on the effect of thrombin on tumor cell lines cultivated in the absence of serum, a potent growth factor.
In addition, the animal data generated from models using serum-cultured transformed tumor cell lines (with unknown chromosomal aberrations) treated with exogenous thrombin before injection do not reflect a true pathophysiologic representation. They ignore endogenous thrombin production/concentration at the tumor-host interface, and it is likely that the concentrations used are at unphysiologic levels and the exposure transient. We therefore elected to study the effect of serum-free thrombin on growth of synchronized tumor cells in vitro as well as the effect of thrombin on spontaneous tumor growth in vivo, using a transgenic TRAMP mouse programmed to develop prostate cancer, a more relevant model.
The inhibitory cell cycle regulator p27Kip1 is an inhibitor of the G1 phase of the cell cycle. It is a negative regulator of protein kinases cyclin-dependent kinase (cdk) 2/cyclin E and cdk2/cyclin A, which drive cells into the S phase. It is an established tumor cell marker of poor prognosis. p27Kip21 is decreased or absent with high tumor grade and poor prognosis with several human tumors ( 20– 28). Its regulation is often at a posttranscriptional level ( 21). The degradation of p27Kip1 is carried out by the ubiquitin system ( 29) that delivers proteins for proteosome degradation. p27Kip21 is degraded by a Skp1-Cyclin-F box protein ubiquitin ligase complex that contains Skp2 as the target recognizing subunit ( 30– 32). Skp2 plays an important role in promoting entry into S phase by inducing p27Kip1 degradation ( 30– 32). Skp2 is commonly overexpressed in numerous human cancers and is an independent prognostic marker ( 33).
MicroRNAs (MiRNA) are small noncoding RNAs that control translation through binding to the 3′-untranslated region of protein coding genes. They inhibit or destabilize translation. Recent data reveal that miRNAs might affect both tumor suppressor and oncogenic potential in human cancer ( 34– 43). It has recently been reported that miRNAs MiR-221 and MiR-222 enhance the proliferation of human prostate LNCaP cells ( 41) as well as glioblastoma, MDAMB-231, and U87 cell lines by down-regulation of p27Kip1 ( 20). We therefore chose to investigate the effect of thrombin on p27Kip1 as well as MiR-221 and MiR-222 in LNCaP cells.
We studied the effect of thrombin on activation of the cell cycle, using starved synchronized tumor cells in the absence of serum. We measured the up-regulation and down-regulation of several cell cycle regulators and focused on p27Kip1. Here, we describe the down-regulation of p27Kip1 by thrombin through up-regulation of Skp2 and miRNA MiR-222, and show this requirement for thrombin-induced cell cycle activation.
Transgenic TRAMP mice were used to determine whether thrombin enhances the spontaneous development of prostate cancer. TRAMP mice express the SV40 Large T Antigen under the control of the prostate-specific probasin promoter, leading to the spontaneous development of prostate cancer and metastasis over a period of 70 to 200 days ( 44). The SV40 Tag acts as an oncoprotein through interaction with p53 and Rb tumor suppressor gene products ( 45– 47), which have been implicated in the development of prostate carcinoma ( 47, 48). These tumors are androgen dependent and regulated with sexual maturity. We monitored the effect of thrombin on in vivo tumor development via two methods. First, we chronically injected the mice with thrombin, i.p. Second, we investigated the loss of endogenous thrombin by chronic injection of hirudin, a highly specific, potent thrombin inhibitor. Here, we describe markedly increased spontaneous prostate cancer growth with thrombin treatment (in the absence of enhanced tumor angiogenesis) and decreased tumor growth with hirudin treatment, indicating that thrombin contributes to in vivo tumor growth.
Materials and Methods
Reagents. All reagents (including human thrombin and androgen-depleted serum) were purchased from Sigma. unless otherwise noted. Hirudin (Refludan) was purchased from Hoechst Marion Roussel. Culture media DMEM was obtained from Mediatech. All vascular growth factor and receptor antibodies (anti–vascular endothelial growth factor, KDR, ANG-2, Tie-2, GRO-1, and CD31) were obtained from Santa Cruz Biotechnology. Anti–cyclin D1 was obtained from Chemicon. Anti-Skp2 were purchased from Zymed Laboratories. Cyclin A is a previously described rabbit polyclonal antibody (C). BrdUrd was obtained from Amersham. Anti–BrdUrd-FITC, MoAb, was purchased from Pharmingen.
Cell lines and culture conditions. The human LNCaP prostate cancer, TRAMP C1, and T98 glioblastoma cells were purchased from American Type Culture Collection.
Mice. Transgenic C57BL/6 female TRAMP mice [heterozygous for the probasin-Tag (SV40) transgene] were a gift from Dr. D. Levy at the NYU School of Medicine, New York, NY. TRAMP mice were crossed with either C57BL/6 or FVB nontransgenic males. These mice have prostatic intraepithelial neoplasia by 12 wk, with tumor arising at 24 wk in the dorsal and lateral lobes of the prostate, appearing as well-differentiated adenocarcinomas. Metastasis to lymph nodes and lung is usually noted at 30 wk. FVB/TRAMP mice give larger tumors, with primary pathology at 12 wk, mostly high-grade. Some well-differentiated prostate cancer metastasis to lymph nodes and lung is generally seen earlier, at 18 wk. Mice were genotyped by PCR using the primers 5′AGGTCTTGAAAGGAGTGCCTGG-3′ and 5′GAGTCAGTAGCCTCATCAC-3′ to give a 654 bp fragment. Mice were injected i.p. at 6 wk with either thrombin (25 units/kg) or hirudin (10 mg/kg) for 10 d daily followed by every other day until sacrifice.
Knock-in and knockdown experiments with p27Kip1 and Skp-2. The p27Kip1 knock in plasmid was a gift from Dr. M. Pagano's laboratory, NYU School of Medicine, New York, NY. The p27Kip1 cDNA encoding protein was subcloned into the EcoR1 site of expression vector pcDNA3 (Invitrogen Life Technologies). See Supplementary Figure S1 for data.
Skp2 shRNA was introduced into the shRNA-RetroQ retrovirus (BD Biosciences; Clontech) at the BamHI and EcoRI ligation sites according to the manufacturer's directions, as we have described ( 49). shRNA Oligonucleotides were derived from the murine Skp2 sequence and synthesized after derivation from the computer program supplied by BD Biosciences.
Forward strand sequence 5′-gatccGGGAGTGACAAAGACTTTGttcaagagaCAAAGTCTTTGTCACTCCCtttttt-3′.
Reverse strand sequence 5′-aattcaaaaaaGGGAGTGACAAAGACTTTGtctcttgaaCAAAGTCTTTGTCACTCCCg-3′.
Scrambled shRNA oligo pairs.
Forward strand sequence 5′-gatccGGCGTAACATGAGGATAGTttcaagagaACTATCCTCATGTTACGCC-ttttttg-3′.
Reverse strand sequence 5′-aattcaaaaaaGGCGTAACATGAGGATAGTtctcttgaaACTATCCTCATGTTACGCCg-3′.
Immunoblots. Immunoblotting was performed as previously described ( 9). Briefly, cells were starved for 72 h in 0.1% bovine serum albumin before repletion with serum or treatment with thrombin. Cells were then lysed in RIPA buffer (20 mmol/L Tris-HCI, 150 mmol/L NaCI, 1% Triton X-100, 5 mmol/L EDTA, 1% deoxycholate, 0.5% SDS, 1 mmol/L Na3VO4, plus 1 μg/mL Aprotin, 1 μg/mL leupeptin, 50 μg/mL phenylmethylsulfonyl fluoride, and 1% NP40) and 25 μg of extract in 25 μL applied to SDS-PAGE. The immunoprecipitate was washed and boiled in SDS sample buffer. After SDS-PAGE and transfer to nitrocellulose membranes, Western blotting was performed with the appropriate antibody, secondary antibody, and enhanced ECL reagents (Supersignal/West Pico; Pierce).
Immunohistochemistry. Ten percent formalin-fixed tumor was embedded in paraffin and stained for von Willebrand factor ( 50) or Ki67 as previously described ( 1, 8). Angiogenesis was evaluated in a blinded fashion by two different observers.
Tumor pathology. Mean tumor area (mm2) was obtained by averaging the areas from 10 of 100 serial slices of tumor (every 10th slice).
Cell cycle analysis. Cells (1 × 106) were serum starved for 72 h in 0.1% fetal bovine serum in DMEM media. Thrombin (0.5 unit/mL) was added for varying time intervals of 0 to 24 h. Cells were then centrifuged, stained with anti–BrdUrd-FITC for 30 min at room temperature, recentrifuged, washed, and stained with propidium iodide before fluorescence-activated cell sorting analysis (with a kit purchased from BD Biosciences).
Northern analysis. LNCaP cells were extracted for RNA with TRizol (Invitrogen). Fifteen micrograms of total RNA were loaded onto a 10% denaturing polyacrylamide gel. Immobilon-Ny + Transfer Membranes (Millipore Corp.) were used for electrotransfer of RNA. Probes were labeled with T4 polynucleotide kinase- [32P]ATP. Hybridization buffer (Clontech) was used for Northern blot hybridization according to the manufacturer's instructions. The probe sequences were as follows: miR-222, 5′-gagacccagtagccagat-3′; U6.5′-cacgaatttgcgtgtcatccttgcgca-3′.
Effect of thrombin on tumor mitogenesis and activation of the cell cycle in human LNCaP prostate and T98 glioblastoma tumor cell lines as well as a murine TRAMP prostate tumor cell line. The potential mitogenic effects of thrombin were examined in three tissue culture systems in serum-starved synchronized tumor cells. First, the effect of thrombin administration on LNCaP prostate cancer cells was analyzed. After 72 hours of androgen starvation, LNCaP cells arrest in the G0 phase of the cell cycle, so we assessed the ability of thrombin to cause arrested LNCaP cells to re-enter the cell cycle using BrdUrd incorporation and propidium iodide staining. Addition of thrombin (0.5 u/mL) had essentially the same effect as addition of androgen-containing serum ( Fig. 1A and B ), inducing cells to leave G0, enter G1, and progress to S phase. By 8 hours posttreatment, the number of S phase cells increased dramatically from baseline for both the serum (29-fold) and thrombin-treated cells (48-fold). We further analyzed the S phase changes between 2 and 24 hours for thrombin or PAR-1 agonist TFLLRN. This increased 2.2-fold (P < 0.04) and 3.5-fold (P < 0.004), respectively (n = 6). To confirm these results, we performed similar studies with TRAMP ( Fig. 1C and D) and glioblastoma cell lines, T98G (Supplementary Fig. S2), which arrest in G0 in response to serum starvation. Similar to the results in LNCaP cells, serum-starved TRAMP cells responded to thrombin as well as its PAR-1 activation peptide, TFLLRN by increasing 2.8-fold for both (P < 0.001; n = 4). T98G cells also entered S phase in response to both serum and thrombin, suggesting that thrombin is a general mitogen. Thus, thrombin per se is a potent stimulator of the cell cycle compared with the resting state.
Examination of cell cycle kinases and inhibitors. We further analyzed the effect of thrombin on the cell cycle by performing immunoblots on cell cycle components known to be modulated during cell growth and proliferation ( Fig. 2A ). In cells synchronized by serum/androgen starvation, levels of p27Kip1, a cyclin-dependent kinase inhibitor, are high, whereas levels of cyclins D1 and A, the activating subunits for cyclin-dependent kinases are low. Confirming the effects of thrombin on these cells, both thrombin and serum addition result in down-regulation of p27Kip1 with the concomitant induction of Skp2, the E3 ubiguitin ligase for p27Kip1. Cyclins D1 and A are induced with similar kinetics by both thrombin and serum, with cyclin D1 up-regulation by 4 to 8 hours and cyclin A by 4 to 18 hours. Quantification of thrombin effect by density measurements is given in Fig. 2B. These results confirm that thrombin has a mitogenic effect on cells, ruling out an in vivo angiogenesis requirement for thrombin-induced tumor growth.
Thrombin activates the cell cycle through the ligation of PAR-1. We next examined the thrombin-induced mechanism of cell cycle activation by using the thrombin PAR-1 activation peptide TFLLRN. Figure 3Aa and Bb show similar TFLLRN activation kinetics as found with pure thrombin, including the down-regulation of p27Kip1 for LNCap cells. We wondered whether p27Kip1 is rate limiting for thrombin activation of the cell cycle. We therefore studied the inhibitory role of p27Kip1 on thrombin-induced cell cycle activation. Transfected p27Kip1 LNCaP cells inhibited the stimulation effect of thrombin, Fig. 3Ac and Ad or TFLLRN ( Fig. 3Ae) on cell cycle activation. The effect of thrombin on p27Kip1 and Skp2 was next examined because Skp2 is involved in the degradation of p27Kip1. Thrombin ( Fig. 3Ba) or TFLLRN-stimulated ( Fig. 3Bb) cell cycle was inhibited in Skp2 KD LNCaP cells ( Fig. 3Bc and Bd).
Thrombin down-regulates p27Kip1 through the up-regulation of MiR-222. MiRNA-222 is up-regulated in numerous cancers, many of which are accompanied by p27Kip1 loss or mutation ( 34, 42, 43). A recent report on prostate cancer cell lines reveals an inverse relationship between p27Kip1 with MiR-222 ( 41). We therefore hypothesized that thrombin up-regulates MiR-222. This hypothesis was tested by Northern analysis of thrombin-activated LNCaP cells with and without p27Kip1 transfection. Figure 4 shows up-regulation of MiRNA-222 in wild-type cells ∼2-fold but not in p27Kip1-transfected cells at 2 and 8 hours (P > 0.1). Thus, we can confirm an inverse relationship between p27Kip1 and MiR-222 and now show that thrombin down-regulation of p27Kip1 requires up-regulation of MiR-222.
Effect of thrombin on spontaneous tumor growth in TRAMP mice. We next designed an in vivo experimental program to test the effect of chronic thrombin treatment on the development of spontaneous tumor growth in TRAMP mice, obviating the effect of in vitro thrombin-tumor culture followed by in vivo injection of a large unphysiologic bolus of tumor. Animals were treated at 6 weeks with 25 units/kg thrombin or 10 mg/kg hirudin/mouse every day for 10 days followed by every other day until age 140 to 175 days, as indicated. Six experiments were performed with five to six animals in each group, using two different genetic backgrounds, C57BL/6 and C57BL/6/FVB. Measurements of tumor incidence, tumor area, tumor size versus whole prostate (% tumor), and mean prostate weight was recorded for each mouse.
Effect of thrombin on prostate cancer and angiogenesis in TRAMP mice. The result of repetitive thrombin treatment is shown in Fig. 5A and Table 1A . Tumor area, tumor percentage and prostate/tumor weight increased 3-fold (P = 0.02), 5.6-fold (P = 0.035), and 7.7-fold (P = 0.028), respectively. The incidence of carcinoma was also greater in thrombin-treated compared with PBS-treated mice (7 of 9 versus 2 of 9, respectively; P = 0.028). Although the angiogenesis index was enhanced 1.64-fold (1.1 ± 0.1 versus 1.8 ± 0.15; P < 0.02) in 9 livers of thrombin-treated animals, no enhancement was noted in 9 tumor specimens (P > 0.1; Supplementary Fig. S3). Because no thrombin enhancement was noted for tumor associated angiogenesis (von Willebrand factor staining) the data support the concept that in vivo thrombin-induced growth is independent of angiogenesis.
Effect of hirudin on the mitotic index and prostate cancer in TRAMP mice. To test the effect of endogenous thrombin, animals were treated with repetitive hirudin (potent antithrombin). To examine the effect of endogenous thrombin on the mitotic index of tumors, we first compared hirudin-treated TRAMP mice with PBS-treated controls using Ki-67 staining to monitor cell proliferation. Figure 5B and Table 1B show the hirudin-induced inhibition of prostate tumor growth in TRAMP of both mouse strains (C57BL/6 and C57BL/6 FVB). Tumor area, tumor percentage, and prostate tumor weight decreased 7.6-fold (P = 0.03), 24-fold (P = 0.03), and 12.6-fold (P = 0.04), respectively. The incidence of carcinoma was also lower in hirudin-treated animals compared with PBS-treated control (4 of 12 versus 7 of 10; P = 0.09). The lack of significance is likely due to a longer time period chosen for observation as well as a more susceptible mouse strain (C57BL/6/FVB). Figure 5C (top two) confirms the beneficial effect of hirudin on tumor infiltration. Note the PBS-treated poorly differentiated prostate tumor growing in a diffuse pattern compared with the hirudin-treated tumor demonstrating early atypical preneoplastic changes with focal glandular hyperplasia, cribiform features, and mild atypia. Figure 5C (bottom) clearly shows inhibition of tumor mitogenesis by hirudin. Positive staining with Ki-67 represents nuclear staining in a proliferating cell. It is conceivable that the hirudin effect was nonspecific. However, we think this unlikely because hirudin had no effect on tumor growth in vitro and hirudin is a highly specific inhibitor of thrombin.
These studies clearly show that thrombin is a growth factor that stimulates spontaneous prostate mitogenesis by inducing activation of the cell cycle from G0-G1 to S by down-regulation of p27Kip1, after activation of PAR-1. Of particular interest is the potent growth stimulating effect of pure thrombin, which was equipotent with serum (a mixture of growth factors). Thrombin down-regulates p27Kip1, a negative regulator of the cell cycle, in association with the up-regulation of Skp2 and MiR-222, which decrease p27Kip1 by different mechanisms. Thus, thrombin operates postranscriptionally by inactivating p27Kip1.
The in vitro cell cycle data were confirmed in vivo with a spontaneous prostate cancer phenotype found in TRAMP mice. These studies, along with the cell cycle in vitro studies, show that thrombin can contribute to tumor growth. Indeed, tumor-associated angiogenesis was not significantly different between thrombin-treated and PBS-treated animals, suggesting that the tumor itself induces sufficient potentially maximal angiogenesis in the TRAMP model. Thrombin-treated animals increased their tumor volume 6- to 8-fold, whereas hirudin (potent-specific inhibitor of endogenous thrombin) decreased tumor volume 13- to 24-fold. The very potent effect of hirudin reflects the importance of endogenous generation of thrombin during spontaneous tumor development. These data are consistent with our previous reports that show the rate-limiting effect of the thrombin PAR-1 receptor on experimental pulmonary metastasis (a highly artificial pathophysiologic situation; ref. 5), as well as the effect of hirudin on external tumor implantation, seeding, and spontaneous metastasis (an artificial implantation situation; ref. 8). Our present data reflect a more relevant spontaneous tumor development situation.
These new observations accentuate the importance of thrombin in spontaneous tumor growth and development—and could be responsible for activation of a dormant tumor state, as suggested in a recent review ( 1).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: NIH grants HL 13336 and DA 04315 and the Hildegarde D. Becher Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Oral Presentation at the Annual Meeting of the American Society of Hematology, San Francisco, December, 2008.
- Received November 11, 2008.
- Revision received February 5, 2009.
- Accepted February 9, 2009.
- ©2009 American Association for Cancer Research.