Previous Article | Next Article ![]()
Molecular and Cellular Biology, February 2006, p. 1452-1462, Vol. 26, No. 4
0270-7306/06/$08.00+0 doi:10.1128/MCB.26.4.1452-1462.2006
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Lynne W. Elmore,2
Kimberly Haydu,2
Colleen K. Jackson-Cook,2,3,4 and
Shawn E. Holt1,2,3,4*
Department of Pharmacology and Toxicology,1 Department of Pathology,2 Department of Human Genetics,3 Massey Cancer Center, Medical College of Virginia at Virginia Commonwealth University, 1101 E. Marshall St., Richmond, Virginia 23298-06624
Received 25 July 2005/ Returned for modification 4 October 2005/ Accepted 18 November 2005
|
|
|---|
|
|
|---|
Because of its nearly ubiquitous expression in human cancer, telomerase is an obvious chemotherapeutic target (40). Telomerase activity requires two core components, hTERT and hTR (10, 28, 46), to be assembled into a functionally active enzyme by the Hsp90 chaperone complex (20). We have previously demonstrated that chaperones are essential for optimal telomerase assembly in vitro (20) and that Hsp90 itself remains associated with the functional telomerase complex (11).
In a human prostate cancer model, increased assembly of telomerase by chaperones, including Hsp90, has been shown to correlate with prostate cancer progression, which is defined as increased aggressiveness in vivo (1). These findings indicate that increased expression of the Hsp90 chaperone complex with the associated activation of telomerase may be important steps in prostate cancer formation (1, 20). While telomerase in cancer progression has been widely studied (reviewed in reference 40), the role of chaperones in carcinogenesis and their interplay between telomerase and its substrate, the telomere, are less well defined.
Many studies indicate that Hsp90 chaperone inhibitors, such as geldanamycin (GA), 17-allylamino-17-demethoxy-geldanamycin (17-AAG), and radicicol (RAD), may be clinically useful as therapeutic agents for cancer patients (reviewed by 29, 13, 19). These inhibitors are capable of simultaneously targeting multiple Hsp90-associated proteins that are important in tumorigenicity, including N-ras, Ki-ras, HER-2, c-Raf-1, Akt, and mutant p53, ultimately resulting in the induction of cytostasis and/or apoptosis in cancer cells (21, 25, 41). Hsp90 is also involved in the production of free radicals from the nitric oxide synthase (NOS) pathway (27, 31, 34). Despite several studies describing the effect of chaperone inhibition on telomerase activity, few studies have examined the long-term consequences of Hsp90 inhibition on telomere length using either pharmacological or genetic approaches. Thus, the goal of our study was to establish a relationship between NOS-induced free radical production and telomere damage after genetic and/or pharmacologic disruption of Hsp90 function.
|
|
|---|
Cell lines and isolation of subclones. All tumor cells were cultured in RPMI 1640 containing 5% fetal bovine serum and supplemented with ITS (insulin, 5 µg/ml; transferrin, 5 µg/ml; and selenium, 5 ng/ml; Collaborative Research), dexamethasone (0.1 µM), and gentamicin (0.05 mg/ml). The human prostate epithelial cell line, M12, used in these studies has been extensively characterized (1-3). All cells were mycoplasma free, as assessed by the mycoplasma T.C. Rapid Detection system (Gen-Probe, San Diego, CA). Isolation of subclones was achieved by seeding 500 cells onto a 15-cm2 tissue culture plate and then expanding individual colonies after 2 weeks.
Design of siRNAs.
Small interfering RNA (siRNA) sequences were designed according to the manufacturer's recommendations for use with the pSUPER.retro (SUPpression of Endogenous RNA) system (Oligoengine). Briefly, a 19-nucleotide target sequence specific to the chaperone of interest was identified using Dharmacon siDESIGN center, and criteria were reviewed as described previously (5). Candidate sequences were used to synthesize a pair of 64-mer oligonucleotides. The complementary oligonucleotides were annealed according to the manufacturer's instructions prior to cloning into the pSUPER.retro vector. The 64-mer siRNA sequences were synthesized for both isoforms of Hsp90 (
and ß) and the
isoform only, as follows: HSP90 (
and ß), 5'-GATCCCCGTTT GAGAACCTCT GCAAATTCAA GAGATTTGCA GAGGTTCTCA AACTTTTTGG AAA-3'; and HSP90
, 5'-AGCTTTTCCA AAAAGTTTGA GAACCTCTGC AAATCTCTTG AATTTGCAGA GGTTCTCAAA CGGG-3'. The presence of the correct insert was confirmed by sequencing using the primers specific for p-SUPER.retro at positions 1242 to 1257 and its complement at positions 2645 to 2629.
Generation of retroviral cell lines and infection. The pSUPER.retro vectors were transfected using Fugene reagent (Roche) into the Phoenix A competent cell line, as recommended. The resulting retroviral supernatant was collected, filtered through a 0.45-µm filter, and incubated with M12 cells. Cells were allowed to recover before selection with 1 µg/ml puromycin, followed by isolation of individual clones as described above.
RAD treatments. For chronic RAD studies, M12 cells were seeded at 25% confluence and exposed to freshly prepared 0.3 µM RAD (1 mM stock in DMSO) or an equivalent volume of DMSO in standard RPMI culture medium. RAD-containing (or vehicle-containing) medium was replaced every 2 days. Cells were cultured and reseeded at 25% confluence, usually on the fourth day. At each passage, total cell numbers were determined, and population doublings were calculated using the formula [log10(number of cells counted/number of cells plated)]/0.3. All cells were kept in log-phase growth during the calculation of proliferation rates.
Terminal deoxynucleotidyltransferase-mediated dUTP-X nick-end labeling (TUNEL) assay. Cells were fixed directly on a six-well chamber slide in 4% formaldehyde for 10 min. Next, the cells were washed twice in phosphate-buffered saline (PBS) for 5 min each and fixed in acetic acid:ethanol (2:1) at 20°C for 5 min. Slides were washed, blocked in 1 mg/ml bovine serum albumin in PBS for 30 min at room temperature, and incubated with an enzyme mix containing 4 µl terminal transferase, 5x reaction buffer, 25 mM CoCl2 and fluorescein 12-dUTP (Boehringer Mannheim) for 60 min at 37°C in a humidified chamber under light-sensitive conditions. Cells were washed twice in PBS for 5 min, mounted in Vectashield (Vector Labs), and stored at 4°C. Representative images were captured using an OLYMPUS IX70 fluorescence microscope (Optical Elements Corporation).
Immunoblotting.
Cells grown in DMSO or radicicol (0.3 µM) or after stable integration of siRNAs were harvested at
80% confluence, and proteins were isolated and electrophoresed as described previously (8). Nitrocellulose membranes were subjected to immunoblotting as described previously (8) with anti-Hsp90, anti-p23, (generous gifts of David Toft, Mayo Clinic, Rochester, MN; all 1:5,000 dilution), antiactin (Sigma) (1), antiubiquitin (Stressgen, Vancouver, British Columbia), or anti-universal NOS (anti-uNOS; Stressgen) (1:1,000), followed by incubation with secondary antibody. Detection involved the use of a Pierce ECL kit with exposure to Kodak X-OMAT film.
Measurements of superoxide generation. Dihydroethidium (DHE) (Sigma) is a cell-permeative dye that is oxidized to fluorescent ethidium bromide by superoxides and intercalates into DNA. Prior to addition of DHE, cells were thoroughly rinsed to remove all traces of drug and then incubated in fresh medium containing 0.5 µM DHE for 30 min at 37°C in the dark. Intact cell images were captured by fluorescence microscopy. Alternatively, cells were treated as described above except that after incubation in dye, cells were trypsinized, washed, and resuspended in PBS at 1 x 106/ml for analysis using flow cytometry.
Telomeric repeat amplification protocol (TRAP assay). Telomerase activity was measured according to the manufacturer's instructions using the TRAP-eze detection kit (Serologicals, Purchase, NY), as described previously (1, 11). Relative telomerase activity was quantitated by ImageQuant software (Molecular Dynamics) analysis of the ratio of the telomerase ladder to the included 36-bp internal standard.
TALA. A telomere amount and length assay (TALA) was modified from the work of Gan and colleagues (12). Genomic DNA was digested using six restriction enzymes, AluI, MspI, RsaI, CfoI, HaeIII, and HinfI (Gibco-BRL). A radiolabeled telomere-specific probe, (TTAGGG)4, was added to digested genomic DNA, followed by hybridization for 2 h at 55°C. Samples were cooled to 4°C, loaded onto a 0.8% agarose gel, and electrophoresed for 18 h. Dried gels were wrapped in plastic and exposed to phosphorimaging cassettes. Average telomere lengths were determined as described previously (32).
Cytogenetic analysis. Standard chromosomal harvest procedures were used with a colcemid time of 2 h (37). Metaphase chromosomes were visualized using conventional Giemsa staining (4), and the chromosomal changes that were observed were tallied. Metaphase spreads (n = 50) were scored for chromosomal findings from the DMSO- and RAD-treated M12 cell lines at day 59 after treatment.
|
|
|---|
Traditional studies using Hsp90 inhibitors utilize an acute treatment, assessing the cellular effects in the short term (29). To determine the long-term consequences of Hsp90 inhibition after chronic exposure to the less-toxic RAD compound, M12 prostate cancer cells were cultured in medium containing a low concentration (0.3 µM) of RAD for 60 days. DNA was isolated periodically from vehicle (DMSO)-treated or RAD-treated M12 cells, and telomere lengths were analyzed. We found that telomeres remained at a constant level ranging between 3.5 and 4.0 kb in the presence of vehicle, while cells treated with RAD undergo a gradual telomere shortening starting at
20 days, continuing throughout the course of treatment (Fig. 1A). Average telomere lengths quantified from multiple assays indicate that telomeres shorten from
3.75 kb to
1.5 kb during chronic RAD treatment, equating to a shortening rate of
200 bp per population doubling. Although chronic treatment of M12 cells at 0.3 µM RAD could be achieved without effects on cell viability/toxicity in the short term, continued treatment resulted in a delayed induction of cell death after
55 days (n = 6; death reproducibly occurred between days 54 and 60) (Fig. 1B). We found that nearly 100% of the treated cells undergo apoptosis as assessed by TUNEL (Fig. 1C). Cells treated with vehicle are TUNEL negative (Fig. 1C), suggesting that critically short telomeres in cells chronically treated with RAD signal the activation of the apoptotic cascade.
![]() View larger version (41K): [in a new window] |
FIG. 1. Treatment of prostate cancer cells with radicicol transiently inhibits telomerase activity and reduces telomere length, culminating in apoptosis. (A) Genomic DNA was isolated from cells exposed to chronic RAD and digested with six restriction enzymes. The digested DNA was hybridized with a 32P-labeled telomere probe, and telomere signals separated by gel electrophoresis. A size ladder in kilobase is noted on the left of the blot. (B) Cells continuously cultured in 0.3 µM RAD or vehicle (DMSO) were assessed for cell growth in log phase over the course of treatment. (C) Following chronic RAD treatment for 55 days, cells were stained with terminal transferase and fluoroscein 12-dUTP (TUNEL) and 4',6'-diamidino-2-phenylindole (not shown). All 4',6'-diamidino-2-phenylindole-positive RAD-treated cells were stained with TUNEL, visualized by an intense fluorescent (green) signal, while the DMSO controls were not TUNEL positive. (D) The M12 prostate cancer cells treated with 0.3 µM radicicol or DMSO were periodically analyzed for telomerase activity (100 cell equivalents) by the TRAP assay. IC is the 36-bp internal PCR control.
|
Inhibition of Hsp90 results in the ubiquitination of a number of its bound target proteins, which are then targeted to the ubiquitin-mediated proteasome degradation pathway (29). To determine if the cellular response to RAD remains constant throughout the course of treatment (i.e., constant Hsp90 inhibition), cells were analyzed for global changes in the levels of ubiquitinated proteins. Chronic RAD treatment increases the levels of ubiquitinated proteins compared to those in cells treated with vehicle (data not shown), and those levels remain constant throughout the course of treatment, indicating that Hsp90 function remains inhibited in RAD-treated cells.
Free radical production in cells chronically treated with radicicol. The observation that telomeres shorten in the presence of detectable telomerase activity may at first appear contradictory to our understanding of how telomerase maintains telomeres. We therefore set out to establish the mechanism of telomere shortening in RAD-treated cells. Interestingly, inhibition of Hsp90 by GA, 17-AAG, and RAD has been implicated in the production of oxidative free radicals by disruption of the NOS pathway (31, 42). Therefore, we asked whether RAD treatment might induce oxidative free radicals that contribute to the telomere shortening.
M12 cells were exposed to various concentrations of RAD for 24 h, washed free of excess drug, and incubated with 5 µM DHE, a reduced form of ethidium dye that is oxidized by reactive oxygen species (ROS) to fluorescent ethidium, which incorporates into DNA. Cells treated with increasing concentrations of RAD have a concentration-dependent increase in ROS levels (Fig. 2A), which indicates that chronic treatment of RAD could lead to an accumulation of DNA damage induced by a constant exposure to oxidative free radicals. We then confirmed that chronic Hsp90 inhibition by RAD caused an increase in ROS generation (Fig. 2C), as reflected by more-intense DNA staining than in the short-term experiments (Fig. 2A).
![]() View larger version (15K): [in a new window] |
FIG. 2. Radicicol induces NOS-dependent free radicals. (A) RAD induces a dose-dependent increase in superoxide (O2) in the M12 cells. Cells were incubated in fresh medium containing DHE for 30 min, followed by imaging using a fluorescence microscope. Cells generating high concentrations of free radicals emit an intense red fluorescence. (B) Hsp90 and Hsp70 were immunoprecipitated using a NOS-specific antibody or with control immunoglobulin G antibodies. Following immunoprecipitation, proteins were transferred to nitrocellulose and immunoblotted for either Hsp90 or Hsp70 to determine association with NOS. (C) Cells chronically treated with RAD or DMSO for 55 days were reseeded into eight-well chamber slides, incubated with DHE, and visualized. (D) Cells were incubated in 0.3 µM or 1 µM RAD for 24 h with or without 500 µM of the NOS inhibitor L-NAME, and then incubated in fresh medium containing DHE. Free radical production was quantified by flow cytometry analysis and plotted on a logarithmic scale.
|
Production of free radicals and telomere shortening are dependent on nitric oxide synthase activity. In previous studies, investigators demonstrate that the production of free radicals by Hsp90 inhibitors originates from the uncoupling of NOS enzymatic function (31). To demonstrate that free radical production in our system was derived from uncoupling of NOS, cells were incubated with 1.0 µM or 0.3 µM RAD for 24 h, followed by 1 h with the NOS inhibitor L-NAME (500 µM). L-NAME completely inhibits production of free radicals and nitric oxide, rendering the NOS enzyme functionally inactive. Simultaneous incubation of RAD with the NOS inhibitor L-NAME caused a shift in the DHE staining intensity as quantified by flow cytometry, corresponding to a reduction in ROS production compared to that with RAD treatment alone (Fig. 2D; also data not shown).
To determine whether the cells (and their telomere lengths) could recover from chronic RAD treatment, cells were cultured for 49 days with 0.3 µM RAD, and on day 50, drug was removed or treatment continued. While the cells maintained in RAD consistently underwent apoptosis between days 55 and 60, as shown in Fig. 1C, cells removed from RAD continued to proliferate for more than 35 doublings beyond this point (Fig. 3A), with growth rates that were not appreciably different from those of DMSO-treated cells.
![]() View larger version (52K): [in a new window] |
FIG. 3. Telomere regrowth and prevention of apoptosis in cells after radicicol is removed. (A) Growth of M12 cells continuously cultured in the presence of RAD for 49 days. At day 50, cells were either maintained in RAD medium (or DMSO) or removed from drug (or vehicle control). Growth was calculated as population doublings from logarithmically dividing cells. (B) RAD was removed at day 49, and cells were cultured for the indicated numbers of days beyond drug (or vehicle) removal. The average telomere length (Q) was measured after the TALA procedure using ImageQuant software as described previously (32), with the size ladder in kilobases to the right of the blot. Note the increase in mean telomere length after RAD removal, as indicated by the quantitation below each lane.
|
To determine if inhibition of NOS would also prevent telomere shortening and delayed apoptosis in RAD-treated cells, cells were treated chronically for 44 days, and on day 45, cells were maintained in RAD (or DMSO) with or without the NOS inhibitor L-NAME. Cells treated with RAD alone died as expected, while cells cocultured with L-NAME continued to proliferate well beyond the point at which RAD-treated cells underwent apoptosis (Fig. 4A). Telomere lengths in control cells (DMSO or DMSO/L-NAME) were similar to lengths in untreated cells. While chronic RAD treatment resulted in very short telomeres, telomere lengths of cells treated with the combination of RAD and L-NAME appeared to be maintained (Fig. 4B). Importantly, telomere lengths did not continue to shorten, suggesting that the shortest telomeres may be maintained in order to prevent cell death. This observation that simultaneous treatment with RAD and NOS inhibitors prevents free radical production and cell death directly implicates NOS as the source of the telomere damage. Thus, it appears that without constant damage of telomeres by free radicals, a functional telomerase is capable of at least maintaining telomeres, thereby preventing apoptosis.
![]() View larger version (44K): [in a new window] |
FIG. 4. Inhibition of the NOS pathway prevents telomere shortening and apoptosis in RAD-treated cells. Cells were cultured for 44 days in RAD medium. On day 45, cells were either maintained in RAD alone (or DMSO) or cocultured in L-NAME (or DMSO) as indicated. Cells were assessed for (A) growth over time and (B) telomere length, determined using TALA. Three plates were simultaneously treated and counted to give an average growth rate. Size ladders are noted to the right of the blot in kilobase pairs (B).
|
Hsp90 exists in two isoforms encoded on separate genes, referred to as Hsp90
and -ß for human cells (35). In humans, Hsp90ß is constitutively expressed and moderately inducible, whereas Hsp90
is generally expressed at low basal levels that are induced dramatically in response to stress (13). For Hsp90 siRNA constructs, target sequences were selected in a region that is conserved in all members of the Hsp90 family. In addition, a similar siRNA was created to specifically target only the Hsp90
isoform. While there is evidence to suggest some overlap in function (33, 48), there are also some Hsp90 isoform-specific functions as well (9).
M12 prostate cancer cells were infected with vector (pSUPER), the Hsp90 siRNA (targets both Hsp90
and -ß), or the Hsp90
siRNA, and expression of Hsp90 and Hsp90
was monitored by immunoblotting. Compared to empty vector controls, those cells stably expressing the
/ß Hsp90 siRNA had a considerable reduction of total Hsp90 protein expression (Fig. 5A), while cells expressing the Hsp90
siRNA showed only a reduction in Hsp90
(Fig. 5B). Single-cell-derived clones were isolated (6 to 12 for each siRNA type), and representative clones with at least an approximately 50% reduction in Hsp90 protein abundance (both for the
/ß Hsp90 siRNA and for the Hsp90
siRNA) were used in the remainder of these studies (Fig. 5C). Since we were unable to isolate individual clones with much less than 50% reduction in Hsp90 expression, this is presumably the minimum level of Hsp90 chaperone activity required to maintain cell survival.
![]() View larger version (41K): [in a new window] |
FIG. 5. Genetic inhibition of Hsp90 causes telomere shortening without altering telomerase activity or cell growth. (A) Mass cultures of M12 cells with siRNA expression specific for either Hsp90 or both Hsp90 isoforms ( and ß), along with vector (pSUPER) controls, were tested for expression of Hsp90 by immunoblotting. (B) Hsp90 protein levels following stable integration of the Hsp90 and Hsp90 siRNA constructs. (C) Hsp90 expression in single-cell-derived clones (1 and 2) after infection with siRNA directed at both Hsp90 isoforms. (D) Growth of logarithmically cultured M12 cells following Hsp90 siRNA expression was calculated as an average for three separate plates, and overall growth was based on three independent experiments. (E) Telomerase activity was tested in M12 mass cultures (left panel) and clones (right panel) following infection with siRNA constructs directed at Hsp90. Five hundred cell equivalents were analyzed for telomerase activity by the TRAP assay. C1/C2 means clone 1 or clone 2. (F) Telomere lengths of individual clones were analyzed using 20 µg of genomic DNA and the TALA protocol as described for Fig. 1. Average telomere length (Q) was measured as described previously (32).
|
constructs did not appear to have an observable change in the total Hsp90 (Hsp90
and -ß) chaperone expression level (Fig. 5A). However, the expression of the Hsp90
isoform was dramatically reduced (Fig. 5B). This observation is not unexpected, since it is an immunoblot for total Hsp90, which makes up 2 to 4% of total cellular protein, and small changes in the weakly expressed Hsp90
isoform were expected to be masked by the predominant Hsp90ß isoform (13). Cells stably infected with siRNA constructs targeting Hsp90 displayed no notable changes in proliferation, indicating that cells can survive with reduced expression of these chaperones without any observable cellular toxicity (Fig. 5D). Furthermore, there were no measurable changes in telomerase activity between populations of cells or individual clone samples that expressed siRNA inserts and empty vector controls (Fig. 5E). Creation of cell lines that stably express siRNA-like constructs requires a 5- to 7-day antibiotic selection process, followed by analysis by immunoblotting and/or TRAP. Therefore, it is likely that a transient inhibition of telomerase activity, like that observed in drug-treated cells between days 2 and 6, would be missed in siRNA cells due to the time frame of the selection process. Taken together, the data from both the drug and genetic inhibition of Hsp90 studies described above leads to the conclusion that there are no long-term effects on telomerase activity in the presence of reduced functional Hsp90.
Telomere length analysis of the siRNA cells confirmed the findings using RAD, indicating that reduction in Hsp90 function, whether global Hsp90 or Hsp90
, leads to telomere shortening (Fig. 5F). The extent of shortening was similar in both Hsp90 siRNA-infected cells and chronic RAD-treated cells, with a reduction from 3.3 kb down to 1.5 to 2 kb (Fig. 5F). It has previously been shown that there is significant clonal heterogeneity in terms of telomerase activity and telomere length in established cell lines (38), with a range of longer and shorter telomeres in single-cell-derived clones. While we show data for only four clones, all of the clones analyzed had shorter telomeres than the parental line (data not shown), indicating that telomere shortening has occurred in the Hsp90 siRNA cell lines.
NOS-dependent free radical production induces telomere shortening after genetic inhibition of Hsp90. The similarity between pharmacological and genetic (siRNA) inhibition of Hsp90 function on growth, telomerase activity, and telomere length points to a common mechanism of action that is responsible for induction of telomere shortening in these cells. Given that we observed increased NOS-dependent free radical production combined with telomere shortening following chronic RAD exposure, cells were similarly evaluated following genetic inhibition of Hsp90 for the production of free radicals. As shown with pharmacological inhibition of Hsp90, genetic knockdown of Hsp90 resulted in elevated ROS production above levels observed in cells with an empty vector (Fig. 6A and 6B, right panel). Production of free radicals is reduced after a 1-h incubation with the NOS inhibitor L-NAME (data not shown) and completely blocked after a 24-h incubation (Fig. 6B, left panel).
![]() View larger version (45K): [in a new window] |
FIG. 6. Inhibition of NOS-induced free radicals prevents telomere erosion. (A) Cells expressing the vector (pSUPER) or siRNAs against total Hsp90 or Hsp90 were incubated in medium containing DHE. Increased superoxide levels are visualized by an intense red fluorescence. (B) Hsp90 siRNA-expressing cells were incubated in fresh medium (or medium containing L-NAME for 24 h) and then with medium containing DHE. Free radical production was analyzed by flow cytometry. (C) Telomere lengths (TALA) were analyzed over time in cells expressing Hsp90 siRNAs and compared to those of the pSUPER controls after incubation with 500 mM L-NAME.
|
DNA damage and telomere dysfunction. To conclusively show that DNA damage does occur in these cells, we determined the level of acquired chromosomal changes present in RAD-treated cells versus vehicle (DMSO)-treated cells, as previously done (8). Figure 7 shows representative metaphase spreads clearly demonstrating numerous telomere-specific chromosomal abnormalities in the RAD-treated cells compared to the DMSO-treated cells, the data for which are summarized in Table 1. The M12 RAD-treated cells presented in a continuum, with the least-affected cells having a small number of end breaks and/or end fusions, progressing to a large number of end fusions/breaks along with interstitial breaks, progressing to massive chromatin disintegration. Based on the significantly increased frequency of end breaks and end fusions, one can safely conclude that there is damage at the telomere in these treated cells.
![]() View larger version (20K): [in a new window] |
FIG. 7. Chromosomal damage and telomere dysfunction induced by inhibition of Hsp90 and production of free radicals. Cells were harvested after 59 days of treatment and subjected to a 2-h colcemid block to assess metaphase chromosomes. Metaphase spreads were evaluated from DMSO-treated (A) and RAD-treated (B) M12 cells after 59 days of treatment. The DMSO cell chromosome numbers ranged from near diploid with few structural abnormalities to near triploid (A), while the RAD-treated cells were predominantly near triploid. The M12-treated cells show many cytogenetic abnormalities related to telomere dysfunction, including end associations/fusions and end breaks as indicated by the arrows.
|
|
View this table: [in a new window] |
TABLE 1. Chromosomal abnormalities after radicicol treatment in M12 prostate cells
|
|
|
|---|
![]() View larger version (20K): [in a new window] |
FIG. 8. Model of Hsp90-mediated regulation of NOS and free radical-induced telomere damage. NOS produces nitric oxide (·NO) as a by-product of the reaction in which L-arginine and oxygen are converted to L-citrulline. Hsp90 association with NOS promotes this conversion and suppresses the pathway by which NOS generates the superoxide (O2) free radicals. Inhibition of Hsp90 function pharmacologically (GA or RAD) or genetically (siRNA) disrupts NOS conformation to inhibit ·NO production and promote O2 generation, which results in the accumulation of telomere damage (shortening), rendering the cell susceptible to apoptosis. L-NAME, by blocking both NOS pathways, protects against telomere damage and apoptosis (not depicted in the model).
|
It is interesting that telomerase activity recovers in cells during chronic Hsp90 inhibition, suggesting telomerase assembly in the absence of Hsp90 function. It may be that Hsp90 is not strictly required for assembly of extractable telomerase but is necessary for telomere elongation, consistent with our previous findings that Hsp90 is associated with the functional telomerase enzyme (11). Alternatively, and perhaps more simply, because telomerase is a low-abundance Hsp90 target, it may also be that there is enough residual functional Hsp90 in cells to fully assemble active telomerase. Recent evidence suggests that the TRAP assay for assessing telomerase activity levels may not be sufficiently sensitive to detect the subtle changes in telomerase activity/function observed by Hsp90 inhibition (26, 44), so that even though extractable telomerase activity is abundant, telomerically functional telomerase may be significantly reduced.
Inhibition of Hsp90 disrupts free radical homeostasis. Increasing evidence indicates that Hsp90 is intimately involved in the production of oxidative damage as a result of its interaction with NOS. Hsp90 functionally associates with NOS, an enzyme involved in converting L-arginine into L-citrulline and nitric oxide (·NO) (27). These studies demonstrate that NOS, in addition to producing ·NO, is also capable of producing O2 free radicals and H2O2 (34). In particular, inhibitors of Hsp90, including GA, 17-AAG, and RAD, can disrupt NOS activity, resulting in an increase in NOS-dependent O2 radicals (31). We show that pharmacologic and genetic inhibition of Hsp90 results in significant free radical production and that a specific NOS inhibitor (L-NAME) blocks ROS production. Taken together, these results clearly demonstrate a direct link between DNA damage induced by inhibition of Hsp90 and deregulation of the NOS pathway.
NOS-induced free radical production damages telomeres.
It is well established that oxidative free radicals can damage DNA, including telomeric DNA. When combined with the finding that RAD induces high levels of ROS, these two observations suggest that the increased production of free radicals by Hsp90 inhibition may contribute to telomerase-independent telomere damage. Furthermore, several lines of evidence suggest that telomeres may act as preferential targets for free radical damage. When telomeric oligonucleotides were exposed to oxidative stress using H2O2 and O2, DNA damage (i.e., cleavage sites and adducts) preferentially occurred at the 5' site of 5'-GGG-3' sequence in an oligonucleotide (22). Fenton reactions (between H2O2 and Fe2+) cause preferential cleavage and strand breaks located at the 5' end of the sequence RGGG in a plasmid containing 81-telomere repeats (17). They propose that telomeric DNA may protect the genome from DNA damage by attracting oxidative damage to noncoding telomeric sequences (17), which seems plausible given the telomere's repetitive, G-rich sequence and its overall structure. These studies together with our current data allow us to predict that the extent of telomeric damage induced by a lack of functional chaperones is greater than the compensatory mechanisms of telomerase, since detectable telomerase activity failed to protect against telomere shortening in cells with inhibited Hsp90. Clearly, our cytogenetic analysis shows significant chromosome damage of the type consistent with telomere dysfunction, suggesting high levels of telomere damage after treatment with RAD. In an attempt to characterize the DNA damage response, we assessed for the binding of
-H2AX and 53BP1 at the telomeres after treatment with RAD, since these proteins have been shown to be first responders to DNA damage at both interstitial and telomeric sites (7). However, we were unable to show a classic DNA damage response in the RAD-treated cells, even though treatment with
-irradiation resulted in classic DNA damage-induced foci (data not shown) (7). Because radicicol is not a DNA damaging agent, since it specifically inhibits Hsp90 function, this is not an unexpected finding.
Interestingly, prostate cells are capable of recovering from RAD even after extensive telomere damage, since withdrawal from RAD restored telomere lengths to levels observed in untreated cells. This, in turn, suppresses the induction of apoptosis in these cells, thus demonstrating a connection between critically short telomeres and induction of apoptosis.
Further support for this model was demonstrated in our system by showing that the NOS inhibitor L-NAME was capable of eliminating free radical production and protecting cells from telomere shortening, directly implicating the source of the telomere damage as NOS-induced free radicals. The fact that telomerase is expressed in both RAD- and siRNA-treated cells further suggests that this is a telomerase-independent telomere-shortening mechanism. Interestingly, both conditions resulted in high levels of NOS-dependent free radical production, which provides convincing evidence that these two processes are directly linked.
The precise mechanism of how telomere shortening contributes to the onset of drug-induced apoptosis is not immediately clear. We speculate that while telomerase is functionally inhibited from its telomere maintenance mechanism, the telomere erosion observed might be due to multiple mechanisms that result in the deprotection of the telomere and hence telomere dysfunction. Disruption of its structure would likely render the telomere more susceptible to free radical-induced telomere damage, and this structural change could be the result of a disruption of telomere binding protein function. Additionally, because many DNA repair proteins are associated with the telomere (7), it is plausible to suggest that inhibition of Hsp90 may prevent these repair proteins from properly functioning, leading to a lack of telomere repair and an accumulation of ROS-induced telomere damage.
Free radicals have been demonstrated to be intimately involved in the induction of apoptosis, yet the mechanism(s) of free radical-induced apoptosis in response to different stimuli is poorly understood. While both free radicals and telomere shortening can induce apoptosis, we do not understand how these biologic processes interconnect. Consistent with our findings, an earlier study has demonstrated that the consequence of ·OH radical-induced apoptosis was telomere erosion (36), indicating a strong correlation between free radical-induced telomere shortening and apoptosis and adding additional support to our model.
Summary. Understanding the complex nature of DNA damage and apoptosis induced by Hsp90 inhibition, both genetically and pharmacologically, will contribute to improved use of Hsp90-inhibitory compounds therapeutically and may help to identify additional agents that are likely to work synergistically to increase specific killing of cancer cells. The work presented here has provided important insights into the consequences of Hsp90 inhibition on telomerase and telomere biology. While we demonstrate telomerase inhibition in prostate cancer cells by chaperone inhibitors in the short term, our data clearly indicate that telomerase inhibition cannot be maintained long term using these compounds, thus revealing that anti-Hsp90 compounds would be ineffective telomerase inhibitors. However, other aspects of Hsp90 inhibition may make these agents potentially useful as adjuvant therapies. Some Hsp90 inhibitors are currently in clinical trials and have been shown to simultaneously target multiple signaling pathways involved in promoting tumorigenicity. Here, we provide additional insights into the complex pharmacology of these agents with respect to the Hsp90 inhibitor radicicol. We demonstrate that RAD is capable of producing free radicals even at low concentrations, which in turn translates into specific free radical-induced telomere damage. Moreover, we find that long-term culture of cells with the Hsp90 inhibitor RAD results in delayed apoptosis as a consequence of critically short telomeres. Our knowledge of how different chemotherapeutic agents may induce DNA damage and apoptosis by free radical mechanisms may potentially lead to better treatment strategies and adjuvant therapies that can prevent cancer development and disease recurrence.
This work was funded by a New Investigator Award from the Department of Defense (DAMD17-02-1-0152) to S.E.H. and by National Institute of Environmental Health Sciences (NIEHS) [R01 ES12074] to C.K.J.-C.
The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of NIEHS or NIH.
Present address: Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, N.C. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»