Prostate cancer is the most common malignancy in men, and patients with metastatic disease have poor outcome even with the most advanced therapeutic approaches. Most cancer therapies target the bulk tumor cells, but may leave intact a small population of tumor-initiating cells (TICs), which are believed to be responsible for the subsequent relapse and metastasis. Using specific surface markers (CD44, integrin α2β1 and CD133), Hoechst 33342 dye exclusion, and holoclone formation, we isolated TICs from a panel of prostate cancer cell lines (DU145, C4-2 and LNCaP). We have found that prostate TICs have significant telomerase activity which is inhibited by imetelstat sodium (GRN163L), a new telomerase antagonist that is currently in Phase I/II clinical trials for several hematological and solid tumor malignancies. Prostate TICs telomeres were of similar average length to the telomeres of the main population of cells and significant telomere shortening was detected in prostate TICs as a result of imetelstat treatment. These findings suggest that telomerase inhibition therapy may be able to efficiently target the prostate TICs in addition to the bulk tumor cells, providing new opportunities for combination therapies.
Early detection combined with androgen depletion therapy significantly reduces morbidity in patients with localized prostate cancer, but for the patients with metastatic disease the therapeutic options are limited.1 The development of castrate- and drug-resistant tumors poses further challenges in the treatment of prostate cancer.2 Therefore, understanding the etiology of prostate cancer may lead to the development of new chemotherapeutic agents, which can circumvent the limitations of current therapies.
The initial tumor formation and subsequent tumor relapse are believed to be caused by small populations of cells, known as tumor-initiating cells (TICs) or cancer stem cells.3, 4 The existence of TICs was suggested by the observation that cancers are composed of heterogeneous cell populations, with different capacities of tumor initiation.5, 6 According to this hypothesis, targeting the TICs may be the only viable method to eliminate the tumor and achieve a significant therapeutic response. Several experimental strategies have been used to identify prostate TICs. One of the most popular strategy uses specific surface markers such as CD44,7–13 integrin α2β1,14, 15 CD13316 or a combination of the above.17–19 A different approach is to isolate side population (SP) cells based on the exclusion of Hoechst 33342 dye.20, 21 Finally, an innovative strategy is based on the hypothesis that only the holoclones (tightly packed round colonies of cells with distinct morphology) are able to re-initiate tumor growth.22, 23
Telomeres are specialized nucleoprotein complexes that protect the ends of linear chromosomes24 and in the vast majority of human tumors telomere lengths are maintained by telomerase.25 Previous studies have shown that almost all prostate carcinomas have detectable telomerase activity,26–30 and there is a direct correlation between the total amount of telomerase and the Gleason score.31, 32 By contrast, in normal prostate tissues, telomerase activity is absent.33 The increased level of telomerase activity almost universally present in carcinomas and the lack of telomerase in most normal tissues make it an attractive target for anticancer therapy.34–40 One of the most efficient telomerase inhibitors is a N3′-P5′ thio-phosphoamidate oligonucleotide antagonist (GRN163, Geron Corporation, Menlo Park, CA), which causes telomerase inhibition and progressive telomere shortening in numerous cancer cell types.40–43 The second generation of GRN163, designated imetelstat sodium (GRN163L), shows increased intracellular uptake, increased telomerase inhibition and telomere shortening in several cancer cell lines.44–47 Imetelstat has now entered early stage clinical trials as single agent for chronic lymphocytic leukemia and multiple myeloma and in combination with standard chemotherapeutics for non-small cell lung cancer and breast cancer.48
We previously hypothesized that telomerase inhibition can efficiently target the TICs,49 but there are few rigorous scientific investigations that study the telomere biology of TICs. It is generally believed, but not well documented, that TICs are telomerase-positive, but little is known about the telomere length of these cells.50 In this study, we set out to investigate if prostate TICs have telomerase activity and if these cells could be efficiently targeted by telomerase inhibitor drugs such as imetelstat. This report demonstrates that prostate TICs have high levels of telomerase activity and that treatment with imetelstat leads to telomerase inhibition and subsequent telomere shortening in the prostate TICs. These results have important therapeutic implications for telomerase inhibitor drugs in prostate cancer therapy.
Material and Methods
The prostate cancer cell line DU145 was maintained in a 4:1 mixture of Dulbecco's modified Eagle's medium and medium 199 supplemented with 10% cosmic calf serum (HyClone, Logan, UT). The PC3, C4-2 and LNCaP prostate cancer cell lines were grown in T-medium (Invitrogen, Carlsbad, CA) supplemented with 5% fetal calf serum (HyClone, Logan, UT). All the cell lines were kept in a humidified incubator with 5% CO2, at 37°C.
Imetelstat (5′-Palm-TAGGGTTAGACAA-NH2-3′) is an oligonucleotide containing a sequence complementary to the hTR template region of telomerase. For short-term telomerase inhibition, the cells were treated with 1 μM drug 72 h prior to TIC isolation. For long-term treatment, leading to telomere shortening, the cells were passaged weekly and treated with 2 μM drug every 3 days.
Isolation of TICs using surface markers
Cells were grown on 15-cm tissue culture dishes (BD Falcon, Bedford, MA) until they became subconfluent, then gently detached using 0.05% Trypsin EDTA (Invitrogen, Carlsbad, CA). After detachment, the total number of cells was determined using a Z1 Coulter Counter (Beckman Coulter, Fullerton, CA) and the cells were resuspended in cold 1× PBS at a density of 1 × 107 cells/100 μl. The following antibodies and dilutions were used: 1:10 integrin alpha 2 (AK7) mouse monoclonal FITC-conjugated antibody (Abcam, Cambridge, MA), 1:10 CD44 (G44-26) mouse monoclonal PE-conjugated antibody (BD Biosciences, San Jose, CA) and 1:10 CD133 (AC133) mouse monoclonal PE-conjugated antibody (Miltenyi Biotec, Auburn, CA). The cells were incubated with the antibodies on ice for 20 min, then washed twice with cold 1× PBS. After washes, the cells were strained through a nylon mesh (70-μm cell strainer, BD Falcon, Bedford, MA) and maintained on ice until FACS analysis. Cell sorting was performed on a Becton-Dickinson FACSAria (BD Biosciences, San Jose, CA). IgG samples were used as negative controls and the positive cells gated out of the living cell population. Tumor-initiating fractions were sorted for both controls and imetelstat treatment groups.
Isolation of SP (Side Population)
The SP protocol was based on Goodell et al.51 The cells (1 × 106/ml) were incubated in warm T-medium with 5% fetal bovine serum containing 5 μg/ml Hoechst 33342 for 1 h at 37°C with occasional mixing. A control sample was incubated with 50 μM verapamil to confirm the nature of the SP. After incubation, the cells were resuspended in cold 1× PBS and propidium iodide was added to a final concentration of 2 μg/ml before FACS analysis. The samples were analyzed on a MoFlo flow cytometer (Beckman Coulter, Fullerton, CA) with UV excitation at 360 nm. The fluorescence was measured with a 670-nm filter and a 405-nm filter.
Isolation of holoclones and spheroid formation assays
Cells were plated low density (500 cells/10-cm dishes) and after 10 days the colonies were counted and holoclones were isolated based on their morphology using small diameter cloning rings. Holoclones are tightly packed colonies of small cells with round morphology; meroclones possess an intermediate phenotype and paraclones have irregular shape and are composed of large, loosely packed cells. The clones were briefly expanded, then harvested for subsequent analysis. For the assessment of clonogenicity in long-term imetelstat-treated cells, we used both serial dilutions in 96-well plates and 10-cm dishes. For the clonogenic spheroid formation assays, the cells were plated on ultra-low attachment dishes (Corning Life Sciences, Lowell, MA) and the spheroids were counted after 10 days of culture.
The telomerase activity was measured using a Telomeric Repeat Amplification Protocol (TRAP) with the TRAPeze kit (Chemicon, Temecula, CA) according to the manufacturer's instructions. The cells were pelleted, lysed in CHAPS buffer (on ice) and after preparing the PCR reactions with cell lysates equivalent to equal number of cells, the telomerase extension products were amplified using a PTC-200 Peltier Thermal Cycler (MJ Research, Waltham, MA). The samples were resolved on a 10% polyacrylamide gel and visualized using a Typhoon Trio Variable Mode Imager (Amersham Biosciences, Piscataway, NJ). The telomerase products (6-bp ladder) and the 36-bp internal control (ITAS) bands were quantified using the AlphaImager 2000 software (Alpha Innotech, San Leandro, CA). The relative telomerase activity (RTA) was calculated as the intensity ratio of the TRAP ladder to that of the ITAS band, and the relative intensity of each sample was normalized to that of the positive control.
Total DNA was extracted from the cancer cells using the DNeasy Blood and Tissue Kit (Qiagen Sciences, MD). Telomere restriction fragment (TRF) analysis was performed as described previously.46 Briefly, 1 μg of total DNA was digested with a mixture of 6 enzymes and separated on an agarose gel. The gel was denatured, dried and neutralized in 1.5 M NaCl and 0.5 M Tris–HCl at pH 8.0. The gel was then hybridized with a 32P-labeled telomeric probe overnight at 42°C. After several washes, the gel was exposed to a Phosphor screen overnight, which was analyzed using a Typhoon Trio Variable Mode Imager (Amersham Biosciences, Piscataway, NJ).
Inhibition of telomerase activity in prostate cancer cell lines by imetelstat leads to telomere shortening
First, we set out to evaluate the effects of imetelstat on the whole population of prostate cancer cells. Figure 1a shows the expected 6-bp TRAP ladder for 4 different prostate cancer cell lines (DU145, PC3, C4-2 and LNCaP), quantified in Figure 1b. All the cell lines used in this study have significant levels of telomerase activity, 3 of them (DU145, C4-2 and LNCaP) have more RTA (relative telomerase activity) when compared to HeLa cells. Treatment with imetelstat leads to efficient telomerase inhibition (Fig. 1c) in a dose-dependent fashion (gel data not shown). Prolonged telomerase inhibition due to imetelstat treatment leads to telomere shortening in all the prostate cancer cell lines analyzed (Fig. 1d). The telomere lengths of the cells used in our experiments vary in average size, from short (LNCaP) to relatively long (DU145), and there is no correlation between telomere length and telomerase activity. If telomerase inhibition (2 μM every 3 days) was maintained until the telomeres became critically short, the cells ceased to proliferate and ultimately died (data not shown). More importantly, there is a correlation between the interval of time required for the onset of apoptosis as a result of telomere shortening (due to imetelstat treatment) and the initial telomere length.
Prostate TICs isolated using established surface markers are telomerase-positive and sensitive to telomerase inhibition by imetelstat
It was previously shown that DU145 CD44+/integrin α2β1hi cancer cells possess traits of tumor stem/progenitor cells and are more proliferative, clonogenic, tumorigenic and metastatic than the CD44+/integrin α2β1low cells.15 We sorted the top 10% cells stained with each antibody, which translated to ∼2% of the total population (Fig. 2a). Equal numbers of CD44hi/integrin α2β1hi cells were collected for the untreated and imetelstat-treated samples and subsequently used for the TRAP assay. The DU145 CD44hi/integrin α2β1hi cells have telomerase levels similar to that of total population, and imetelstat inhibition of telomerase activity is equally efficient in this putative stem/progenitor cell fraction (Fig. 2b).
In the LNCaP cell line, the isolated CD44+/CD24− population is highly tumorigenic and expresses specific genes known to be important in stem cell maintenance.12 Consistent with this previous study, the percentage of LNCaP cells that stained positive for CD44 was relatively small, less than 1% (Fig. 2c). The LNCaP CD44+/CD24− cells had high levels of telomerase activity, which was similar to the main population of LNCaP cells. Again, imetelstat was able to robustly inhibit telomerase activity in both fractions (Fig. 2d).
CD133+ cells isolated from the DU145 line have the capacity of self-renewal and differentiation, as well as high proliferative and tumorigenic potential.19 An antibody against CD133 (prominin-1) was used to sort a small population of CD133+ cells from the DU145 line (Fig. 3a). DU145 CD133+ cells have high levels of telomerase activity, similar to the levels found in the total population of cells, and imetelstat is effective at inhibiting telomerase in these putative stem-like cells (Fig. 3b).
These results show that populations of prostate TICs isolated using most of the surface markers cited in the literature are telomerase-positive and are sensitive to the telomerase inhibitor imetelstat.
The SP (side population) cells have high levels of telomerase activity which is inhibited by imetelstat
The SP, believed to harbor the TICs, can be isolated by sorting cells which exclude the Hoechst 33342 dye.21, 52 Out of the 4 cell lines analyzed, we detected a small SP only in the C4-2 prostate cell line (an LNCaP derivative), which accounted for ∼0.1% of the total population (Fig. 3c). TRAP assays show that the SP cells isolated from the C4-2 cells have high levels of telomerase activity, slightly higher than the main population of cells, and the enzyme's activity was inhibited efficiently by imetelstat (Fig. 3d).
The average telomere length of sorted prostate TICs is similar to the telomere size found in the main population of cells
Once we established that the prostate TICs have significant levels of telomerase activity and that telomerase inhibitors can target efficiently these cells, we investigated the average telomere lengths of prostate TIC fraction. One possibility was that TICs may have longer telomeres compared to the bulk population of cells. As seen in Figure 4, telomeres of prostate TICs are generally of similar average size with the main population of cells from which they were isolated. While the average telomere lengths in TICs versus main population do not vary significantly, the distribution of the telomere length might be different. This is important because the various methods of TICs isolation currently used may not identify cells with identical phenotype. As seen in Figure 1d, the DU145 cells had 2 relative distinct populations of cells, one with long telomeres (∼6.5 kb), which produce the most intense smear on the TRF gel, and the other with shorter telomeres (∼3.4 kb), which produce a smear of lower intensity. In Figure 4b, for the untreated control, only the larger size subpopulation of telomeres was visible due to the low amount of DNA loaded, but for the CD44hi/integrin α2β1hi cells, both populations of telomeres were visible at similar intensities. By contrast, for the DU145 CD133+ TICs, this phenomenon was not apparent; only the large fraction of telomeres was visible on the TRF gel.
The similar average telomere size of TICs and main population of cells would predict that telomere attrition should occur at equal rates in these cells. To verify this hypothesis, we isolated TICs from cell cultures treated with imetelstat for longer periods of time. As illustrated in Figure 4, the telomerase inhibitor was able to induce telomere shortening in these cells, and it appears that the rate of telomere shortening in these cells is similar to that found in the main population of cells.
DU145 prostate cancer holoclones have significant levels of telomerase activity and relatively short telomeres
It has been reported that out of the 3 types of clone morphology (holoclone, meroclone and paraclone) formed by some prostate cancer cell lines, only the cells which form holoclones (tightly packed small cells) are capable of extensive proliferation and tumor initiation in immunocompromised mice.22, 23 By contrast, meroclones (intermediate phenotype) and especially paraclones (loosely packed large cells) have reduced proliferation and tumorigenic potential. Holoclones are enriched in TICs, and isolation of these populations of cells does not require the use of surface markers in conjunction with FACS or magnetic beads. We used very stringent criteria to isolate several DU145 holoclones (Fig. 5a) and expanded them briefly in culture, just long enough to harvest enough material for the subsequent assays. As illustrated in Figure 5b, telomerase activity in these holoclones was on average only slightly lower than that in the DU145 total population. Moreover, the telomere lengths of 7 DU145 holoclones were shorter than the main population of cells (Fig. 5c), suggesting that these cells will be susceptible to telomerase inhibition treatment.
Long-term treatment with imetelstat may reduce the number of TICs and lead to a decreased capacity of self-renewal in prostate cancer cell lines
When we compared the FACS profile of cells treated for different periods of time with imetelstat, we observed that the proportion of DU145 CD44hi/integrin α2β1hi cells present in the total population decreased proportional with the length of imetelstat treatment (Fig. 6a). Using identical gating criteria on the FACS plots, it was evident that the number of CD44hi/integrin α2β1hi cells decreased dramatically after prolonged imetelstat treatment. This phenomenon was observed only after some telomere shortening occurred, because short-term imetelstat-treated cells have the same percentage of CD44hi/integrin α2β1hi cells as in the untreated population (data not shown). To further investigate if imetelstat treatment gradually eliminates TICs from the population, we proceeded to examine the capacity of imetelstat-treated cells to generate holoclones. As illustrated in Figure 6b, after ∼100 days of treatment, we were unable to detect any holoclone formation when the cells were plated at low density. As expected, the numbers of paraclones increased, probably due to the cells with the shortest telomeres in the population entering replicative senescence, but using our scoring criteria no significant differences in the number of meroclones were observed.7
Because the spheroid formation assay is an indicator of self-renewal capacity, we examined whether long-term treatment with imetelstat has an effect on the capacity of self-renewal at the population level. The clonogenic spheroid formation assay showed that prolonged treatment with imetelstat leads to decreased spheroid formation ability in all of the cell lines analyzed (Fig. 6c), supporting the hypothesis that the TICs are gradually eliminated from the population.
Telomerase expression (limitless proliferation) is one of the hallmarks of cancer, and several experimental therapeutic approaches have focused on exploiting this almost universal characteristic of tumor cells. The major aim of this research was to investigate the effects of telomerase inhibition on prostate TICs also referred to as cancer stem cells, using the telomerase antagonist imetelstat (GRN163L). The first question we sought to answer was if prostate TICs have telomerase activity. We chose several cell lines that vary in their androgen responsiveness, tumorigenicity, telomerase activity and telomere lengths. We used subpopulations of cells that demonstrated the highest capacity of tumor initiation (TICs) according to several published studies.12, 15, 19, 21, 23 The methods used to isolate prostate TICs employed established surface markers (CD44, integrin α2β1, CD133), the ability to exclude Hoechst 33342 dye, or the capacity to form holoclones with tumor initiation potential. The results of our experiments show that prostate TICs have significant levels of telomerase activity as measured by the TRAP (telomerase activity) assay. The telomerase activity of TICs was similar not only to the main population of cells, but also to the TIC-negative fractions (Supporting Information Fig. 1). In this study, we chose to focus on the comparison between the TIC fractions and the main population of cells, because the populations of cells that are negative for the cancer stem cell markers used in this study are not capable of tumor initiation and therefore are of little therapeutic significance. This is important, because the reactivation of this enzyme in the TIC compartment is still a topic of debate. One hypothesis, based on normal stem cell biology, was that TICs may be more quiescent than the majority of rapidly dividing cells in the tumor mass. We previously reported that human tumor cells made quiescent by removal of growth factors downregulated telomerase activity.53 If the prediction that TICs are quiescent was correct, then telomerase activity in TICs would be absent or present at very low levels, similar to normal stem cells. However, based on the available data we cannot completely exclude the possibility that TICs possess significant levels of telomerase activity, despite their quiescent status, and this hypothesis raises interesting questions regarding the role of telomerase in these cells. The present study clearly documents that TICs have significant levels of telomerase activity, similar to the main tumor cell population. This supports the hypothesis that TICs are actively proliferating cells with typical cancer telomerase activity. Importantly, because TICs have similar levels of telomerase to the main population of cells, we hypothesized that treatment with the telomerase inhibitor imetelstat would have similar effects in this compartment. The experimental data in the present study show that imetelstat treatment efficiently inhibits telomerase activity in all the prostate TICs populations analyzed.
Because there is not always a direct correlation between telomerase activity and average telomere lengths in various cultured prostate cancer cells, it was important to determine the average telomere length of TICs. We found that the telomeres of these cells were approximately the same length or shorter than the average telomere size found in the main population; therefore, we assumed that telomerase inhibitors currently being tested in clinical trials will induce telomere attrition in the rare populations of TICs with equal efficiency as the bulk tumor cells. We were able to show that the telomerase inhibition by imetelstat induced telomere shortening in the TIC compartment and postulate that prolonged telomerase inhibition will lead to apoptosis and cell death of TICs, similar to the main population of cells.
Investigating the telomere length of prostate TICs was important, since it was theoretically possible for these cells to have longer telomeres, similar to normal stem cells. The reduced telomere length in the tumor-initiating compartment may also shed some light on the process of malignant transformation in prostate. Telomere shortening can be detected as early as prostatic intraepithelial neoplasia (PIN) and is restricted to the luminal compartment.54, 55 This suggests that the TICs originate from a subset of transient amplifying cells which under chronic inflammation pressure and genomic instability caused by short telomeres reactivate telomerase and after additional mutations lead to prostate cancer.
Another important observation we made relates to the effects of prolonged telomerase inhibition on the fraction of TICs present in the population. The experimental data shows that long-term treatment with imetelstat leads to a decrease in the number of DU145 CD44hi/integrin α2β1hi cells present in the total population of cells due to a reduction in the fluorescence of the whole population. Interesting, the CD133+ cells did not show the same trend (data not shown). Prolonged treatment with imetelstat also correlated with a decreased capacity of DU145 cells to form holoclones. Because one of the main characteristics of TICs is their capacity of self-renewal, it was important to investigate the impact of long-term treatment with imetelstat on the capacity of cells to form spheroids when plated at clonal density in attachment-independent conditions, which was used as a measure of self-renewal capacity for prostate cancer cells.56 When we performed clonogenic spheroid formation assays on long-term imetelstat-treated LNCaP, C4-2 and DU145 cells, telomere shortening positively correlated with a decrease in the sphere-forming ability of these cells, indicating a decreased capacity of self-renewal. This is important, because it was well documented by our group and others57 that telomere shortening is associated with a decreased tumor formation ability in immunocompromised mice.
These experiments support the hypothesis that long-term telomerase inhibition by imetelstat coupled with telomere shortening may lead to the elimination of certain populations of prostate TICs. Whether this is a direct result of the elimination of TICs from the population remains unclear and future experiments will be aimed at answering this important question.
One of the ongoing concerns about telomerase inhibition therapy is related to the effects of long-term telomerase inhibition on normal cells. However, normal prostate cells lack telomerase activity and have longer telomere lengths compared to cancer cells.31, 58 Moreover, we have shown that treatment with imetelstat has no effect on the proliferation of normal cells.46 Normal stem cells are known to be relatively quiescent and have low or no telomerase activity except when dividing. Most importantly, normal stem cells possess relatively long telomeres,59 and we predict that the effect of imetelstat on these normal stem cells would be less toxic in comparison to the shorter telomere length of TICs. Thus, there may be an optimal therapeutic window that would lead to cancer cell death without irreversibly affecting the normal cells.
Telomerase inhibition as single agent therapy is believed to be most effective only after critical telomere erosion occurred, and this may require relatively long periods of treatment (depending on the initial average telomere length of the tumor) to achieve a reduction in tumor mass. In contrast, conventional therapies (such as surgery, radiation and chemotherapy) lead to a dramatic reduction in tumor burden relatively quickly, but do not lead to durable responses in advanced stage cancers. This may be due to the inherent resistance of TICs to conventional therapeutic agents, behavior strongly documented in glioblastoma.60–62 We believe that an ideal prostate cancer therapy should combine conventional therapeutic approaches with telomerase inhibitors, such as imetelstat. While conventional approaches will initially target the bulk tumor, after a certain interval of time imetelstat-mediated telomerase inhibition will shorten the telomeres in the tumor-initiating compartment to a critical level, inducing cell apoptosis and death in this small fraction of cells. This therapy approach could potentially lead to durable responses (Fig. 6).
In summary, this preclinical study shows that telomerase inhibition has a great potential for the treatment of prostate cancer and may be able to target the TICs that contribute to relapse and metastasis.
We would like to thank Geron Corporation (Menlo Park, CA) for providing the imetelstat telomerase inhibitor, Ms. Erin Kitten for technical support and Ms. Angela Diehl for the graphic design. This work was supported by a Department of Defense Prostate Cancer Training Award (Grant no. PC074128 to C.O.M.) and by the Southland Financial Corporation to W.E.W. and J.W.S.