|Home | About | Journals | Submit | Contact Us | Français|
Escape of prostate cancer (PCa) cells from ionizing radiation–induced (IR-induced) killing leads to disease progression and cancer relapse. The influence of sphingolipids, such as ceramide and its metabolite sphingosine 1-phosphate, on signal transduction pathways under cell stress is important to survival adaptation responses. In this study, we demonstrate that ceramide-deacylating enzyme acid ceramidase (AC) was preferentially upregulated in irradiated PCa cells. Radiation-induced AC gene transactivation by activator protein 1 (AP-1) binding on the proximal promoter was sensitive to inhibition of de novo ceramide biosynthesis, as demonstrated by promoter reporter and ChIP-qPCR analyses. Our data indicate that a protective feedback mechanism mitigates the apoptotic effect of IR-induced ceramide generation. We found that deregulation of c-Jun induced marked radiosensitization in vivo and in vitro, which was rescued by ectopic AC overexpression. AC overexpression in PCa clonogens that survived a fractionated 80-Gy IR course was associated with increased radioresistance and proliferation, suggesting a role for AC in radiotherapy failure and relapse. Immunohistochemical analysis of human PCa tissues revealed higher levels of AC after radiotherapy failure than those in therapy-naive PCa, prostatic intraepithelial neoplasia, or benign tissues. Addition of an AC inhibitor to an animal model of xenograft irradiation produced radiosensitization and prevention of relapse. These data indicate that AC is a potentially tractable target for adjuvant radiotherapy.
Over the past decade, with the advent of advanced CT-based treatment, planning intensity modulated radiotherapy has gained ascendency over other radiation approaches for primary prostate cancer (PCa) treatment (1–4). For patients who have not undergone prostatectomy, radiation therapy involves a treatment course of greater than 70 Gy usually administered in daily fractions of 1.8 to 2 Gy over a 7- to 9-week period. A recent study found distant (≤10 years) biochemical control in high-risk patients to be as low as 52.7%, with overall local and distant recurrence rates among all risk groups at 5.1% and 8.6%, respectively (4), similar to previous data (5–10). Although the delivery of higher doses of ionizing radiation (IR) improves local control (11–13), conventional techniques of dose escalation come up against dose-limiting toxicities to noncancerous tissues (4, 14–16). Therefore, for purposes of better control of such patients, the molecular mechanisms underlying PCa cell radioresistance and methods to interdict such resistance must be understood in order to maximize the curative potential of radiation therapy.
Bioactive sphingolipids, particularly ceramide, sphingosine, and sphingosine 1-phosphate (S1P), known as the “ceramide-S1P rheostat” (17), are recognized as critical signaling initiators that regulate cell survival, death, proliferation, and inflammation. As appreciation grows for the role of sphingolipids in vital biological processes (18, 19), efforts to target their expression for therapeutic benefit have also gained traction (20–22). In the context of radiation therapy, characterization of IR-induced sphingolipid processing in programmed cell death has demonstrated ceramide generation through both membrane-associated sphingomyelin hydrolysis and genotoxicity-associated de novo mechanisms (23–27). Stress-activated protein kinase (28) and Bcl-2 family–induced mitochondrial depolarization pathways (25) are proximal downstream targets of ceramide accumulation after IR. However, radioresistance may be elicited by either defects in ceramide generation (29–32) or rapid turnover of ceramide into S1P (33–35). Rescue of the apoptotic phenotype by restoring ceramide accumulation or limiting S1P signaling is currently being studied both at the basic science and clinical levels (36–38).
Irradiation of tumors is a potent death-inducing stimulus that rapidly evolving cancer cells frequently escape by virtue of previously existing mutations in death pathways or by responding to the insult reactively to activate survival pathways. While the characterization of aberrant, cancer-associated gene expression in tissues obtained for diagnosis versus noncancerous tissues is a prominent arena of research (39), the response of tumors to therapy also represents a critical avenue of investigation (22). Work by this group and others has demonstrated that the ceramide-metabolizing enzyme, acid ceramidase (AC), can play an important role in resistance to anticancer therapies (40–47), including IR (26, 36, 48, 49). In this report, we evaluated transcriptional activation of AC in PCa cells treated with radiation. We found that the AC gene (N-acylsphingosine amidohydrolase [ASAH1]) promoter is a target of ceramide-dependent c-Jun/activator protein 1 (c-Jun/AP-1) binding, suggestive of a transactivation feedback trigger to reverse the proapoptotic effects of IR-induced ceramide accumulation. Fractioned irradiation of PCa cells to a cumulative dose of 80 Gy induced and/or selected for PCa cell clonogens, of which 90% demonstrated constitutive AC overexpression that correlated with increased proliferation and radioresistance. AC expression was found to be upregulated in human prostate adenocarcinomas compared with that in noncancer tissues and to an even greater extent in tissues from patients who failed radiation therapy. Finally, we found that AC can be specifically targeted in vivo, in combination with IR, to achieve a durable cure in mouse xenograft models. Taken together, these results indicate that sphingolipid metabolism, particularly ceramide processing by AC, is a functionally important and druggable target to improve responses to radiation therapy.
In order to determine whether IR exposure induces a change in PCa cell sphingolipid expression, we assessed ceramide, sphingosine, and S1P levels in the hormone-refractory PPC-1 cell line treated with 5-Gy IR versus mock irradiation. As shown in Figure Figure1A,1A, cellular sphingolipid levels were not significantly altered between 15 minutes and 8 hours after irradiation (specific ceramide species are shown in Supplemental Figure 1A; supplemental material available online with this article; doi: 10.1172/JCI64791DS1). However, sphingolipid levels were elevated by 12 hours after IR and achieved maximum expression between 24 and 72 hours. Specifically, total ceramide content increased between 40% to 100%, whereas the products of ceramide catabolism, sphingosine and S1P, demonstrated 200%- to 270%-increased expression. Irradiation was also associated with a 3-fold increase of S1P detected in the cell culture supernatants (Supplemental Figure 1B), indicating that IR exposure elicited a marked shift of sphingolipid content toward the soluble products of ceramide metabolism.
Given that several ceramidase isozymes catalyze the hydrolysis of ceramide to sphingosine and free fatty acid, we sought to determine which enzymes were induced by IR based on their pH optima. Homogenates of 5 Gy– and mock-irradiated PPC-1 cells were tested for ceramidase activity at pH 4.5, 7.5, or 9.5 (Figure (Figure1B).1B). Two-fold enhancement of enzyme activity was detected in irradiated cells at acidic pH, while activity at neutral and alkaline pH remained unchanged, suggesting that the lysosomal isozyme, AC, may be largely responsible for ceramide-to-sphingosine processing under irradiative stress. Marked and sustained increase of mRNA expression of the AC gene, ASAH1, but not neutral ceramidase, ASAH2, or alkaline ceramidases, ACER1, ACER2, and ACER3, was detected in PPC-1 cells through 72 hours after IR exposure (Figure (Figure1C).1C). Firefly luciferase–based activity reporter assay demonstrated dose-dependent transcription activation beginning at 12 hours after IR (Figure (Figure1,1, D and E), which corresponded with AC protein upregulation (Figure (Figure1F).1F). IR also upregulated AC in the prostate adenocarcinoma cell lines DU 145, CWR-22Rv1, and LNCaP (Figure (Figure1G).1G). Interestingly, we have not evinced similar upregulation of ceramide content or AC expression by IR exposure to noncancer prostate cell lines PWR-1E (immortalized epithelium), WPMY-1 (immortalized stroma), or PrEC (primary epithelium), as shown in Supplemental Figure 8. Yet, cell lines derived from various other adenocarcinomas or squamous cell carcinomas also demonstrated AC protein overexpression upon γ irradiation (Supplemental Figure 1C). Taken together, these data suggest that many cancers respond to IR by upregulating AC, likely as a response to mitigate the deleterious effects of stress-induced ceramide accumulation.
Based on the above results, we hypothesized that IR-induced ceramide generation constitutes a feedback stimulus that induces AC expression. To test this, we first assessed PPC-1 cell expression of AC under direct ceramide modulation. Exogenous delivery of C6-ceramide, an artificial short acyl-chain ceramide, increased intracellular levels of natural-length ceramides (Figure (Figure2A)2A) and concomitantly enhanced both reporter activity of the ASAH1 promoter (Figure (Figure2B)2B) and expression of AC protein (Figure (Figure2C).2C). Since ceramide profiling indicated that treatment with either IR (50) or short-chain ceramide stimulates a relative increase of C16-ceramide among all species (Supplemental Figure 2A), we ectopically expressed an adenoviral transgene encoding ceramide synthase-6 (CerS6) to preferentially synthesize C16-ceramide. Similar to exogenous C6-ceramide treatment, endogenous ceramide generation by CerS6 significantly enhanced cellular ceramide content (Supplemental Figure 2, A and B), ASAH1 promoter activity (Supplemental Figure 2C), and AC protein expression (Supplemental Figure 2D). These results demonstrated that exogenously or ectopically upregulated ceramide stimulated AC gene expression.
In order to evaluate native mechanisms of ceramide biosynthesis in PCa cell response to IR, we administered inhibitors of ceramide generation prior to γ irradiation. Pretreatment of PPC-1 cells with either an inhibitor of ceramide synthase, fumonisin B1 (FB1) (51), or an inhibitor of serine palmitoyltransferase, myriocin (52), resulted in both suppression of baseline intracellular ceramide levels and abrogation of IR-induced ceramide (Figure (Figure2D).2D). In addition, use of either inhibitor abrogated IR-induced ASAH1 reporter activity (Figure (Figure2E)2E) and AC protein upregulation (Figure (Figure2F).2F). Inhibition of ceramide generation by acid sphingomyelinase, either through desipramine (53) administration (Supplemental Figure 3A) or by RNAi transfection (Supplemental Figure 3B), did not interrupt IR-induced AC upregulation. Altogether, these results implicated the myriocin- and FB1-sensitive pathways of ceramide generation as the primary feedback cues for IR-induced AC upregulation, whereas ceramide generation through sphingomyelin metabolism was not a significant contribution. Furthermore, neither ASAH1 reporter activity nor AC protein expression were altered under conditions of exogenous S1P delivery (data not shown) or sphingosine kinase inhibitor (SKI-II; 1.0 μM) administration prior to γ irradiation (Supplemental Figure 3, C–E), indicating that IR-induced AC transactivation is regulated at the level of ceramide accumulation rather than downstream metabolism into a S1P-dependent feed-forward mechanism.
We next sought to investigate whether this mechanism of AC induction was regulated at the transcriptional level, as the ASAH1 promoter reporter indicated. We subjected nuclear fractions of PPC-1 cells treated with either IR (5 Gy, 24 hours) or C6-ceramide (2 μM, 12 hours) to a DNA-protein binding assay, which demonstrated increased AP-1 activity under either condition (Figure (Figure3A).3A). Targeting the prototypical Jun/Fos members of the AP-1 family with pooled- or single-sequence RNAi transfection implicated c-Jun, and possible c-Fos, involvement in AC gene expression under both basal conditions (Supplemental Figure 4 and Supplemental Figure 5, A and B) and IR induction (Supplemental Figure 5, A and B).
In silico identification of AP-1 consensus binding sequences in the ASAH1 promoter revealed 4 high-density regions of putative binding sites (Figure (Figure3B),3B), which became the targets for interrogation by ChIP. ChIP assay confirmed c-Jun/AP-1 binding within the –628/–418–bp sequence upstream of ASAH1 start codon (Figure (Figure3B).3B). Evaluation of a series of truncated ASAH1 promoter-LUC reporters (54, 55), demonstrated regulatory cis-element(s) involved in IR-mediated AC gene expression within the –496-bp promoter region (Supplemental Figure 5C). TCA→TTG mutations to nullify 3 AP-1 consensus binding sequences within the –496-bp region abrogated IR-induced ASAH1 promoter activation at the site –475/–468 bp, with possible combinatorial involvement of the nearby –494/–486–bp binding site (Supplemental Figure 5D).
Finally, ChIP-qPCR analyses of irradiated PPC-1 cells demonstrated enrichment of c-Jun binding to the ASAH1 promoter (Figure (Figure3,3, C and D). Moreover, this phenomenon occurred in a myriocin- and FB1-sensitive manner. c-Fos binding on the ASAH1 promoter demonstrated a similar trend of upregulation upon IR exposure, which was countered by FB1 and myriocin pretreatment, although results were shy of significance (Figure (Figure3,3, C and E). Interestingly, exogenous C6-ceramide treatment recapitulated induction of c-Jun and c-Fos binding, which was inhibited by FB1 pretreatment but not myriocin pretreatment (Supplemental Figure 6, C and D). Although JunB binding to the ASAH1 promoter was confirmed by ChIP assay (Figure (Figure3C3C and Supplemental Figure 6C), it failed to respond to stimulation with IR (Figure (Figure3C)3C) or exogenous ceramide (Supplemental Figure 6C). Taken together, these results confirmed that c-Jun activity is a predominant factor in the feedback mechanism by which IR-induced ceramide causes ASAH1 transcription.
The functional consequences of IR-induced c-Jun regulating AC was of importance to investigate. IR dose corresponded to increased ceramide expression levels in several PCa cell lines (Figure (Figure4A).4A). Moreover, ceramide levels after IR correlated with clonogenic cell death in the case of each cell line tested (Figure (Figure4B).4B). PPC-1 cells pretreated with inhibitors of ceramide-generating enzymes, FB1, or myriocin were less susceptible to IR-induced caspase-3/7 activation (Supplemental Figure 3F), indicating involvement of the de novo ceramide generation pathway in cell killing. We found that inhibition of AC by RNAi contributes, in large part, to this clonogenic collapse by increasing PCa susceptibility to IR-induced caspase activation, annexin V expression, and apoptosis (Supplemental Figure 10, A–C, respectively). In this study, IR-induced necrosis (Supplemental Figure 10D) was not found to play a significant role in the functional outcome of IR treatment on PCa cells. Importantly, as described below, the rescue of AC overexpression upon IR challenge confirmed its cell survival role in PCa.
In order to assess the contribution of c-Jun–regulated AC to the response of PCa cells to IR, we stably expressed a dominant-negative transactivation mutant of c-Jun, TAM67 (56–58), in PPC-1 cells. Irradiation of these TAM67-expressing cells did not upregulate AC (Figure (Figure4C)4C) and induced even more ceramide accumulation (Figure (Figure4D)4D) than control (pcDNA3 empty vector). AC overexpression stimulated by C6-ceramide administration was similarly attenuated in TAM67-expressing PPC-1 cells (Supplemental Figure 7A). Although TAM67 expression did not change the proliferation rate of PPC-1 cells under basal growth conditions (Supplemental Figure 7B), it did enhance IR-induced clonogenic cell killing compared with empty vector expression or nontransfected parental control (Figure (Figure4E).4E). While this observation implicated a vital role for c-Jun in the cell response to IR, we interrogated downstream involvement of AC by rescuing this phenotype with transduction of AdASAH1 overexpression plasmid (Supplemental Figure 7C). PPC-1–TAM67 cells transduced with AdASAH1-GFP (10 MOI) prior to irradiation demonstrated significant reversal of the TAM67 radiosensitization phenotype (Figure (Figure4E).4E). These results confirm that interference of IR-induced AC transactivation renders PCa cells susceptible to the proapoptotic influence of ceramide accumulation in vitro.
A study of PPC-1–derived xenografts in athymic nu/nu mice elucidated the effects of c-Jun/AP-1 deregulation and ectopic AC reconstitution on PCa resistance to and relapses following radiation therapy (Figure (Figure5,5, C–E). PPC-1–pcDNA3 or PPC-1–TAM67 cells were inoculated into the flanks of nude mice to establish subcutaneous xenografts (128 ± 3 mm3) that were subjected to a hypofractionated schedule of focal irradiation (6 × 5 Gy). There was no significant difference in the growth rate of mock-irradiated pcDNA3 or TAM67 xenografts. However, TAM67 xenografts demonstrated immediate response to radiation therapy, and, upon completion of radiotherapy by day 22, TAM67 xenograft volumes were diminished 81% versus 67% diminution of pcDNA xenografts (P = 0.012). pcDNA3 xenografts demonstrated relapse/regrowth within 8 weeks of therapy initiation, whereas 30 Gy total IR was sufficient to produce cure of 80% of TAM67 xenografts through 35 weeks of evaluation.
In addition to the pcDNA3-TAM67 construct, these PPC-1–derived xenografts also harbored a pcDNA5-tetOASAH1 construct (59) that restored AC overexpression in a doxycycline-responsive manner (Figure (Figure5A).5A). Similar to our observation from adenoviral-based ASAH1 transgene expression, doxycycline induction of AC significantly increased radiation resistance of PPC-1–TAM67 cells in vitro (Figure (Figure5B).5B). In the xenograft model, addition of doxycycline to the drinking water (250 mg/l) of mice bearing TAM67-tetOASAH1 xenografts resulted in increased radiation resistance of the xenografts: upon completion of radiotherapy by day 22, there was no significant reduction of volumes since the beginning of treatment (Figure (Figure5,5, C and D). Furthermore, xenograft regrowth occurred 3 weeks sooner and overall survival of mice was diminished compared with that of irradiated PPC-1–pcDNA3 xenografts (Figure (Figure5,5, D and E). Taken together, these results demonstrate a significant role for c-Jun–mediated AC upregulation in the survival response of PCa cells during radiation therapy.
To examine the association between AC and PCa cell regrowth/relapse after radiotherapy, we administered to PPC-1 cells a regimen of daily 2-Gy fractions to a cumulative IR dose of 80 Gy in vitro. Throughout the course of this experiment, irradiated cells demonstrated enhanced ASAH1 promoter activity (Figure (Figure6A)6A) and AC protein expression (Figure (Figure6B)6B) concomitant with IR exposure. Despite achieving prototypical logarithmic cell killing, we recovered surviving adherent cells at the culmination of radiation treatment and seeded them in single-cell cultures to evaluate clonogenicity. 41%–91% of surviving cancer cells (from the cell lines PPC-1, DU 145, and 22Rv1) maintained clonogenicity after 80 Gy. PPC-1–pcDNA3 cells reconstituted clonal populations at a frequency of 72%, similar to PPC-1 parental cells, whereas only 11% of IR-surviving PPC-1–TAM67 cells reestablished colonies (Figure (Figure6D).6D). Moreover, a majority of the subclones grown from IR-surviving PPC-1, DU 145, and 22Rv1 PCa cell lines continued to overexpress AC up to 8 weeks after the cessation of radiotherapy — 80%, 89%, and 64% respectively — while only 22% of PPC-1–TAM67 subclones overexpressed AC (Figure (Figure6,6, C and D). Interestingly, the levels of AC expression in IR-surviving PPC-1 subclones, quantified by densitometry and normalized to GAPDH, correlated inversely with intracellular ceramide levels (Figure (Figure6E)6E) and correlated positively with radiation resistance (Figure (Figure6F)6F) and cell proliferation rate (Figure (Figure6G).6G). These results pointed to a significant role for AC in the phenotype of aggressive cancer cells that survive radiation therapy and repopulate the tumor.
We then formed the hypothesis that human prostate tissues of patients who have undergone and failed definitive radiation therapy may demonstrate higher levels of AC expression. Indeed, when examining the AC protein expression of archival prostate tissues from patients who suffered biochemical failure (n = 8), we found a significantly higher level of AC expression than in therapy-naive adenocarcinoma (n = 30; P = 0.006), prostatic intraepithelial neoplasia (PIN; n = 26; P < 0.001), or benign prostate tissues (n = 27; P < 0.001; Figure Figure6,6, H–L).These analyses also included 3 pairs of patient-matched PCa tissues before radiation therapy (Figure (Figure6K)6K) and after diagnosis of failure (Figure (Figure6L),6L), which demonstrated increased AC expression (Figure (Figure6H).6H). Taken together, these data not only confirm AC overexpression in frank prostatic adenocarcinoma, as observed previously by immunoblot (60) and RT-PCR (61), but also suggest that further enhancement of expression in response to therapy proffers a selective advantage to PCa cells under irradiative stress. Therefore, it was of keen interest to determine whether AC is a tractable target for combination therapy to improve PCa treatment with IR.
To test the hypothesis that AC ablation improves IR response in PCa cells, we transduced PPC-1 cells with adenoviral ASAH1 shRNA targeting the 3′ UTR (AdshASAH13′UTR-GFP, 5 MOI) prior to γ irradiation. IR-induced AC upregulation was abrogated in cells transduced with AdshASAH13′UTR-GFP but not scrambled-sequence vector control (AdshSCR-GFP; Figure Figure7A).7A). shASAH1 expression resulted in enhanced accumulation of IR-induced ceramide (Figure (Figure7B)7B) and diminished S1P concentration in the cell culture supernatant (Figure (Figure7C).7C). Moreover, shASAH1 transduction into PPC-1, 22Rv1, DU 145, and LNCaP cell lines significantly enhanced susceptibility in each case to IR-mediated clonogenic cell death (Figure (Figure7D7D and Supplemental Figure 9). These results recapitulate our previous observations using transient transfection of siRNA against ASAH1 to sensitize PCa cells to IR (49). Moreover, we transduced another shRNA construct, targeted toward the ASAH1 ORF, and found similarly effective radiosensitization (Figure (Figure7D).7D). Reversal of the radiosensitization phenotype, with transgenic expression of AdshASAH13′UTR-GFP, was attempted with transgenic expression of ASAH1, with or without the 3′ UTR (AdASAH1whole and AdASAH1-3′UTR, respectively; 10 MOI). Although AdASAH1whole expression did not reverse the sensitized phenotype, expression of the mutant AdASAH1-3′UTR restored the baseline response of PPC-1 cells to IR.
We demonstrated PCa cell sensitization to cytotoxic therapy and combination chemoradiation therapy via RNAi ablation of ASAH1 in the context of paclitaxel (Taxol) administration as well. Measurement of ceramide species and sphingosine metabolites in PPC-1 cells exposed to paclitaxel (5 nM for 24 hours) demonstrated marked long-chain ceramide upregulation, without inducing significant changes of very-long-chain ceramide species, sphingosine, or S1P levels (Supplemental Figure 11A). These data suggested a sphingolipid profile shift toward a proapoptotic phenotype, as seen in Supplemental Figure 11D. Apoptosis induction by taxane therapy was abrogated by combination with IR exposure regardless of chemoradiation administration sequence (Supplemental Figure 11C), indicating not only subadditive effect, but also possible antagonism between these anticancer modalities. However, pretreatment of cells with siRNA against ASAH1 yielded increased PCa cell sensitivity to chemotherapy (Supplemental Figure 11, B and D). Moreover, the suppression of ASAH1 expression produced optimal PCa cell sensitization to chemoradiation combination therapy (Supplemental Figure 11D).
Finally, small molecule inhibition of AC in vitro and in vivo was evaluated using a novel lysosomotropic prodrug inhibitor of AC, LCL521 (based on structural modification of the AC inhibitor B13; refs. 43, 62). LCL521 pretreatment of irradiated PPC-1 cells abrogated AC upregulation (Figure (Figure8A).8A). Singular treatment of prostate cell lines demonstrated susceptibility to LCL521 in the low micromolar range (Figure (Figure8B),8B), and combined modality treatment with γ irradiation demonstrated supra-additive PPC-1 cell killing in vitro (Figure (Figure8C).8C). Double-ester bond hydrolysis of the prodrug LCL521 into the active B13 was confirmed by LC/MS analyses of PPC-1–derived xenografts from nu/nu mice 6 hours after LCL521 treatment (intraperitoneal injection at 75 mg/kg; Figure Figure8D).8D). Athymic nu/nu mice bearing subcutaneous PPC-1 cell xenografts (123 ± 4 mm3) were given intraperitoneal LCL521 (75 mg/kg), or cremophor EL vehicle control, prior to charged particle irradiation under a schedule of fractionated irradiation (6 × 5 Gy; Figure Figure8E).8E). Although LCL521 monotherapy produced no therapeutic benefit at this dose, marked response to the combination of LCL521 and IR was realized, compared with cremophor-infused (vehicle control) or untreated subjects. As shown in Figure Figure5,5, the cumulative dose of 30-Gy IR was insufficient to prevent xenograft relapse/regrowth. However, complete response was observed with IR in combination with LCL521 (Figure (Figure8,8, E and F). Taken together, these results demonstrate that AC is a clinically relevant and druggable target for enhanced radiotherapeutic control of PCa.
This study demonstrates that AC upregulation is a conserved response to radiation therapy across multiple tumor types and AC inhibition can directly improve the clinical response to radiotherapy in vitro and in vivo. IR exposure induced AC upregulation at the mRNA, protein, and enzyme activity level. Dependence of this process on ceramide expression was confirmed by exogenously or endogenously manipulating ceramide levels as well as suppression of IR-stimulated ceramide biogenesis. This observation led to our hypothesis that a critical threshold of intracellular ceramide exists beyond which a mechanism of AC gene transactivation is evoked in order to temper ceramide-mediated cell death. A transcription factor activity array and subsequent RNAi validation revealed that the c-Jun/AP-1 transcription factor significantly induces upregulation of AC following IR through de novo ceramide generation. In our hands, 90 percent of surviving clonal cell populations after 80-Gy IR overexpressed AC, and the level of AC expression correlated positively with proliferation and subsequent radiation insensitivity. Human prostate tissues, in addition to exhibiting AC upregulation in the tumor compared with benign tissues or PIN lesions, had significantly higher levels of AC following radiation therapy failure. In murine xenograft models, interference of AC gene transcription by the TAM67 dominant-negative mutant of c-Jun, RNAi against the AC gene, or lysosomotropic small molecule inhibition of the AC enzyme improved the therapeutic response to IR. Importantly, experiments with combination IR and LCL521 therapy provided a durable cure following completion of therapy. Finally, we demonstrated by in vitro clonogenic studies that rescue of AC overexpression in cells with endogenous AC suppression restored radioresistance and relapse following IR therapy proving the importance of this enzyme in the resistance pathway.
Curiously, the AC upregulation induced by radiation exposure in numerous adenocarcinoma and squamous cell carcinoma cell lines was not recapitulated in 3 noncancer prostate cell lines we evaluated: primary prostate epithelium (PrEC), immortalized prostate epithelium (PWR-1E), and immortalized prostate stroma cells (WPMY-1) (as seen in Supplemental Figure 8). Profiling the ceramide species of irradiated (5 Gy) and mock-irradiated cells revealed differences in total ceramide expression and relative species levels between the noncancer and cancer cell lines. These results suggest associations between a 2-fold increase of the overall intracellular ceramide burden, and/or specifically among the C16-ceramide species, and AC induction. Investigations to distinguish differences in sphingolipid profile expression between “normal” versus malignant cells at baseline or in vitro growth conditions have identified higher levels of ceramide expression in cancer tissues and cells as well as species-specific variations (63, 64). At the level of enzymatic function, elucidating the control of sphingolipids upon chemotherapeutic or radiotherapeutic challenge may yield novel differences between cell types (65–70). This offers the tantalizing possibility that a distinct difference in sphingolipid regulation at clinically relevant IR dosage presents targets for therapeutic enhancement. Work in this laboratory is currently invested in identifying the underpinnings of such differential responses.
The mechanisms and pathways of irradiation-induced cell death in PCa cells are critical avenues of investigation. Canonical pathways of cell death would include the well-characterized pathway of de novo ceramide accumulation subsequent to irradiation supported by the findings of the Kolesnick group and others (23, 38, 51, 71, 72). The Kolesnick group characterized this de novo ceramide generation pathway in HeLa cells as dependent on CerS5/6 activation (70). Initial observations indicating that similar ceramide generation pathways were functioning in the PCa cells included the delayed and sustained onset of ceramide accumulation consistent with de novo ceramide synthesis rather than induction of ceramide accumulation through the action of sphingomyelinase. Ceramide accumulation was positively correlated with radiation dose and sensitivity to radiation, as evaluated by clonogenic assay. Furthermore, inhibition of both ceramide synthases with FB1 (51, 72) and the de novo pathway with myriocin (52) prevented IR-induced AC upregulation, ceramide generation, and IR-induced caspase activity, whereas inhibition of acid sphingomyelinase had no effect. Thus, we appear to be observing the diminution of proapoptotic ceramide signaling by ceramide-mediated induction of AC, an enzyme with demonstrated oncogenic activity in PCa (42), which in turn directly catabolizes ceramide itself.
We sought to further characterize the pathways of cell death in response to irradiation in PCa cells. Radiation treatment of cells induces marked G2/M arrest, with increased apoptotic cells in a dose- and time-dependent manner, as determined by annexin V staining. Caspase-3 activity was increased following irradiation, with activity augmented by siRNA inhibition of AC. Apoptotic cell death mediated by de novo ceramide accumulation critically functions in mediating cell death following irradiation. Inhibition of necrotic cell death did not rescue IR-induced cell death in PPC-1 cells, suggesting that necrosis, unlike apoptosis, does not appear to be the major cell death pathway in this setting. Although we identified the cell death pathway following IR through de novo ceramide generation and apoptosis augmented by the inhibition of AC, we have not yet ruled out a contribution of mitotic catastrophe mediated by factors such as caspase-2. This mechanism of IR-induced cell damage/death is of great interest to us, as ceramide may be of importance to this process (73).
Although the onset of ceramide accumulation would predict a favorable outcome to therapy, due to ceramide’s roles in apoptosis and growth arrest, we also observed substantial increases in the ceramide catabolite sphingosine and its anabolite S1P. As free sphingosine is mainly produced by deacylation of ceramide, this observation was highly suggestive of increased flux through 1 or more of the 5 human ceramidases. Interestingly, we found that AC was increased specifically, with no significant alteration in activity or mRNA expression of neutral or alkaline ceramidases. This may be due, in part, to the increase of C16-ceramide — a preferred substrate of AC (44) — relative to other species under conditions radiation exposure or exogenous ceramide treatment. We might also posit that the wave of autophagic activity engendered by a dose of IR that is not immediately lethal accounts for a conserved induction of lysosomal components (74, 75), of which AC is a significant member. How this process may be connected to IR-stimulated c-Jun/AP-1 activity (76, 77) remains underappreciated (78) and an important matter for investigation. In addition, no significant change of sphingosine kinase activity in irradiated PPC-1/PC-3 cells (49, 79) coincides with IR-induced AC, which led to our hypothesis that AC represents a rate-limiting step for IR-induced ceramide catabolism.
Ceramide is a pleiotropic signaling lipid and IR induces myriad alterations to cellular signaling (as reviewed in ref. 80). Mechanistically, we were interested to find that inhibition of ceramide generation following IR by pretreatment with FB1 or myriocin prevented not only ceramide accumulation, but also upregulation of AC transcription and protein expression. This observation lead to the hypothesis that ceramide generation induces AC upregulation as a feedback mechanism to dampen the apoptotic ceramide signal. Moreover, the rapid catabolism of stress-induced ceramide into mitogenic/oncogenic S1P counteracts apoptosis signaling propagated by ceramide accumulation (17). Intriguingly, we observed that nonmalignant prostate epithelial and stromal cell lines do not upregulate AC with IR exposure, which may reveal a survival mechanism that contributes to evasion of apoptosis in malignant cells, as opposed to the orderly apoptotic program maintained by normal cells.
A role for ceramide in regulating gene expression has long been appreciated (81–87). In order to seek out the transcriptional mechanism for IR-induced AC upregulation, we identified transcription factor alterations that were common to ceramide treatment and IR individually. Both ceramide and IR activated AP-1, and siRNA knockdown of the various Jun/Fos components indicated c-Jun involvement in these processes. c-Fos involvement was implicated as well, although results were shy of significance and, therefore, equivocal. Interestingly, JUNB binding was confirmed at the ASAH1 promoter as well, although the level of interaction between this protein and the promoter was unaffected by IR or C6-ceramide induction. Also of note was the failure of myriocin to inhibit c-Jun or c-Fos binding stimulated by C6-ceramide. These observations may result from the bypass of de novo ceramide generation achieved with the exogenous administration of a ceramide source, which may feed the recycling pathway of ceramide biosynthesis, mediated by FB1-sensitive ceramide synthases (88).
The c-Jun proto-oncogene has long been appreciated for mediating cellular transformation, growth, and angiogenesis (89–95) and may have more labile activity in malignant cells (96). The Kester group demonstrated that AP-1 transactivates neutral ceramidase in human coronary artery smooth muscle cells with stimulation by fetal bovine serum (97). Here, through a combination of luciferase reporter truncations/mutations and ChIP–RT-PCR, we deduced the node of c-Jun/AP-1 interaction with the ASAH1 promoter to the –475/–468–bp site. It was suggested that AP-1 control of neutral ceramidase transcription is tightly regulated by predicted overlapping binding of NF-Y or interaction with proximate cis-binding factors, considerations that may direct context-dependent transcription of either neutral or AC under disparate stimuli. Investigation of other transcription factors regulating AC expression, such as CREB (54) and KLF6 (98), via RNAi studies have not yet revealed involvement of those particular factors in IR-induced AC upregulation (data not shown). However, it should be noted that in silico detection by TFBind of CREB, Stat-1, and Elk-1 consensus binding sequences on the same promoter region as AP-1 may suggest competition for or interaction of binding on the AC gene promoter under various conditions of stimulation.
Approximately 50% of patients with cancer undergo radiation therapy (99, 100), typically administered in fractionated doses up to 2 Gy every weekday for 7 to 8 weeks in order to avoid toxicity to adjacent structures (101). However, experimental models of fractionated radiation to clinically achievable doses remain underreported. In the present study, we attempted to mimic clinical irradiation of cancer cells, irradiating PCa cells in vitro at 2-Gy doses for 40 times for the cumulative IR dose of 80 Gy. In our analysis of the surviving clones, we were surprised to find that fractionated irradiation did not completely ablate the reproductive capacity of cancer cells, and, interestingly, 90% of these IR subclones continued to overexpress AC. Although neither the average radiation resistance nor proliferation rate of these IR subclones was as robust as therapy-naive cells, their levels of AC expression showed remarkable correlation with these parameters, harkening to the eventual biochemical failure and relapse of a proportion of radiotherapy patients. As seen in our animal studies using PPC-1 xenografts bearing a doxycycline-inducible AC expression cassette, irradiated mice with forced AC overexpression demonstrated a poorer response to IR compared with mice without AC induction. The irradiated mice overexpressing AC also exhibited earlier relapse compared with IR-treated mice with no AC induction, suggesting a significant contribution to therapy resistance. Taken together, these results demonstrate that AC supports a radioresistance phenotype and inform the hypothesis that enhanced AC expression would be found in prostate tumors that fail definitive courses of radiation therapy.
This was confirmed by evaluation of human prostate tissues by immunohistochemistry. These data suggest that AC is upregulated following IR and support the notion that AC may contribute to resistance and relapse. Retrospective analysis of tissues from difficult to obtain relapsed malignancies showed that radiotherapy patient samples have elevated AC expression. These data provide an indication that our in vitro interrogation of PCa cell responses to IR revealed valid findings. Future studies could use quantification of AC protein in human plasma, urine, or expressed prostatic secretions, which are amenable to high-throughput platforms and allow noninvasive evaluation of tumor response to IR and potential for relapse. This is an attractive approach to studying AC in cancer cells, as their enhanced autophagy and secretory activity enriches the extracellular concentration of lysosomal contents (102–104) and because AC can be detected and isolated from human urine (105).
Clinically achievable radiation doses are limited by systemic and local toxicity to adjacent structures. Thus, adjuvant therapy that selectively sensitizes tumors to IR might enable lower doses with lower adverse events. The efficacy of IR on xenografts with tightly controlled AC expression supports the notion of inhibiting AC as a strategy for improved cancer therapy. Indeed, our experiments indicate that targeting AC upregulation by interfering with c-Jun activation and by small molecule inhibition augments the efficacy of IR without an increase in total dose of radiation (Figures (Figures44 and and5).5). Moreover, inhibition of AC with the novel lysosomotropic prodrug AC inhibitor, LCL521, prevented relapse (resumption of tumor growth after apparent eradication) in the mice (n = 10), whereas approximately 70% of mice receiving only IR experienced relapse. This situation recapitulates relapse after definitive radiation therapy in human patients with locally advanced disease, 50% of who ultimately experience recurrence of disease. These promising in vivo studies suggest that AC may be an ideal clinical target in the setting of solid tumor irradiation.
An important concern with radiosensitizing agents is the potential for sensitization of normal tissue or augmentation of systemic toxicity, effectively defeating the purpose of adjuvant therapy. Our studies indicate no increased toxicity in mice receiving IR plus LCL521 in comparison with that in the group receiving only LCL521. Although no benefit was observed for LCL521 as a single agent against tumor growth at the dose administrated, synergistic cell killing in combination with IR characterizes LCL521 as an efficacious radiosensitizer. Using our rationally designed AC inhibitors, a previous study demonstrated that susceptibility to AC inhibition corresponds to the level of basal AC expression in treated cells (106). This suggests that AC inhibition can produce greater therapeutic benefit in treated tumors, due to specific AC induction via focal IR delivery. Additional evaluation of radiosensitization of nontargeted tissue is necessary as we move forward with evaluation of this therapeutic modality; however, the specificity and lack of apparent unintended sensitization is encouraging. Additionally, while LCL521 is at this point highly experimental and its clinical feasibility is not well known (but currently under evaluation), our study clearly outlines a role for AC inhibition in radiosensitization of prostate tumors.
In summary, our data strongly support AC as a major factor in the development of radiation resistance in PCa therapy. Multiple methods clearly demonstrate that interference of AC induction and activity sensitizes cancer cells to IR in vitro and in vivo. Upregulation of AC in radiotherapy-relapsed human tumors confirms the described phenomenon and justifies further study into the feasibility and efficacy of targeting AC in patients receiving IR for PCa. Therefore, AC inhibition presents a rational approach for improving radiotherapy in the preclinical environment with the potential for advancement into patients.
The human prostate cell lines PPC-1 (gift of Yi Lu, University of Tennessee, Knoxville, Tennessee, USA), DU 145 (ATCC), PC-3 (ATCC), and CWR22Rv1 (ATCC) were maintained in RPMI 1640 media supplemented with 10% bovine growth serum and incubated in 5% CO2 at 37°C. The LNCaP progression series of cell lines was a gift from Leland W.K. Chung (Cedars-Sinai Medical Center, Los Angeles, California, USA) and was cultured according to published methods (107, 108). Cell lines were validated in by STR profiling using the PowerPlex1.2 marker set. Enumeration of cells in any experiment was performed using a Scepter automated cell counter (Millipore), followed by confirmation of viability through Trypan blue exclusion. FB1 (Enzo Life Sciences no. BML-SL220) was dissolved in double-distilled H2O to a concentration of 25 mM. Myriocin (Enzo Life Sciences no. BML-SL226) was dissolved in DMSO (to 10 mM). The MUSC Lipidomics Synthetic Core Facility provided C6-ceramide, which was dissolved in DMSO (to 25 mM), and LCL521, which was dissolved in ethanol (to 50 mM). Solvent/vehicle administrations to cell cultures were conduct at dilutions of 1:1,000 or higher. Additional information is provided in the Supplemental Methods. Primer and RNAi nucleotide sequences are listed in Supplemental Tables 1 and 2.
Cell cultures were irradiated in a137Cs γ-irradiator (J.L. Shepherd & Associates) at an average dose rate of 2.15 Gy per minute based on decay, as previously described (49). Dosimetry using radiochromic (self-developing) film was performed regularly to verify accuracy of delivery.
Athymic nu/nu Mus musculus (NCI Frederick) between 7 and 12 weeks of age were inoculated subcutaneously with 5 × 106 cells in a 1:1 RPMI 1640/Matrigel HC admixture (0.08 ml per injection). Animals were enrolled into studies upon establishment of flank xenografts >100 mm3. Doxycycline administration consisted of feeding animals 250 mg/l doxycycline in 2% sucrose water, which was refreshed every 2 days. Combined modality therapy with LCL521 administration constituted intraperitoneal delivery of drug dissolved in 1:2 cremophor EL/saline solution to a dose of 75 mg/kg per animal. Two to four hours after injection, animals were subjected to sham or focal charged-particle irradiation delivered with a 9-MeV electron beam generated by a Varian 2100EX linear accelerator with dosing prescribed to a depth of 1 cm. Mice were anesthetized and positioned in individual slots of a custom-designed scaffold. A 5-mm-thick custom lead shield was used to protect the normal tissue of the mice. Circular windows of either 1.25-cm or 1.75-cm diameters drilled into the lead shield exposed 1.0-cm or 1.5-cm diameter tumors, respectively. The mice were placed 100 cm from the radiation source, and a 1.0-cm tissue-equivalent bolus (Superflab) was placed on top of the tumors to provide sufficient dose build up. Dosimetry was verified using radiochromic (self-developing) film prior to irradiation. Tumor volumes were measured every 48 hours with digital calipers (Fisher) and recorded as volume (mm3) = [4π/3] × length × width × depth. Animal weight was monitored, and tumor xenografts growing larger than 2,000 mm3 or developing ulcers were considered as having reached the end point; all other events removing animals from the studies were considered censored events.
Ceramides species, sphingosine, and S1P from cell pellets, culture supernatants, and xenograft homogenates were collected and analyzed with LC-MS/MS by the Lipidomics Shared Resource, MUSC, as previously described (49). Synthesis of the AC inhibitors, LCL521, LCL522, and B13, was conducted in the Lipidomics Shared Resource.
Adenoviral vectors expressing CMV-driven CerS6 were designed by Christina Voelkl-Johnson and Tejas Tirodkar and were manufactured by Vector Biolabs. They also provided anti-CerS6 antibody (Abnova). CMV-driven AdASAH1whole-GFP and AdASAH1-3′UTR-GFP and U6-driven AdshASAH13′UTR-GFP and AdshASAH1ORF-GFP were developed by Open Biosystems and Vector Biolabs. CMV-driven Ad-GFP (no. 1060) and U6-driven Ad-shScr-GFP (no. 1122) control adenoviral vectors were obtained from Vector Biolabs.
Immunoblot analyses of cell lysates were performed as previously described (49) using antibodies to detect AC (BD Transduction no. 612302), CerS6 (gift of C. Voelkl-Johnson), β-actin (Sigma-Aldrich no. A5441), and GAPDH (Santa Cruz Biotechnology, no. sc-32233).
Protein-DNA complexes were isolated using a Nuclear Extraction Kit (Affymetrix no. AY2002) and evaluated by the Protein/DNA Array I according to manufacturer’s instructions (Affymetrix no. MA1210). Protein-bound DNA probes are hybridized to the array and assessed by chemiluminescence. Signal intensity data are quantified by densitometry using Quantity One software (Bio-Rad).
We cloned 1.5 kb of the human ASAH1 promoter (NCBI accession AF220172) using the following primers: forward 5′-GATCCTCGAGCTCCACTGCATTTGTCAC-3′, reverse 5′-GATCTGATCAATCGCTCTAGCAGCCAAC-3′. The amplified product was ligated into the XhoI (5′) and BamHI (3′) cloning site of the pLVX-Luc vector (Clontech no. 632162) occupied by the PTight promoter. Stable expression of the reporter construct, pASAH1-Luc, or the control pLVX-Luc vector was established by culturing transfectants under puromycin selection conditions per manufacturer’s instructions. Firefly luciferase activity was measured with the Bright-Flo Luciferase assay kit (Promega) in 96-well microtiter plates using 104 cells per well. Luminosity was measured by a FLUOstar Optima plate reader (BMG), and data shown represent the integration of 10 minutes of luciferase readings during the plateau phase of luciferase activity.
Clonogenic survival was assessed as previously described (49). Briefly, cells in each treatment group were plated in triplicate into 35-mm culture plates after IR. After 10 to 14 days of culture, cells were fixed in 3.7% formaldehyde and stained in 1.0% crystal violet and 0.5% acetic acid, whereupon colonies of ≥50 cells were enumerated. Data representing geometric mean ± SEM were fitted to the linear-quadratic model, and the mean inactivation dose was calculated according to Fertil et al. (111).
Paraffin-embedded formalin-fixed human prostate specimens were evaluated for AC protein expression as previously described (102). Briefly, anti-AC primary antibody (1:100; Santa Cruz Biotechnology, sc-100646) or IgG3 control was diluted with DAKO antibody diluent, added to the slides, and incubated at 4°C overnight. A biotinylated link antibody along with a streptavidin biotin peroxidase kit (DakoLSABþ System-HRP) was then used along with a DAB chromogen and peroxide substrate to detect the bound antibody complexes. The slides were counterstained with hematoxylin and dehydrated through graded alcohols to xylene. A pathologist, blinded to tissue identity, scored 3 areas per tissue core and evaluated them using the following scale: 0, no staining of any cells; 1, faint staining; 2, moderate intensity staining; and 3–5, intense staining.
Unless otherwise indicated, data represent mean ± SEM of 3 independent experiments (n = 3) and were tested for statistical significance by 1-way ANOVA and Tukey-Kramer post-hoc analysis, assuming α = 0.05, with GraphPadInstat 3. Pairwise comparisons of only 2 data groups were subjected to nonparametric analysis of statistical significance (Mann-Whitney U test). For xenograft experiments, differences in average fold change between groups were evaluated on the log scale using 2 sample t tests, with a P value of less than 0.05 considered statistically significant.
Archived, deidentified human tissues were obtained and evaluated under the auspices of Medical University of South Carolina Institutional Review Board protocols HR20613 and NHR397. Animal care, handling, and experiments were conducted under the auspices of Medical University of South Carolina Institutional Animal Care and Use Committee protocol AR1575.
This work was funded by NCI P01CA097132-07 (to J.S. Norris) and NCRR UL1 RR029882 (South Carolina Clinical and Translational Research Institute). Support to the Lipidomics Core Facility, Biorepository and Research Pathology Services, and Biostatistics and Clinical Trials Shared Resource was provided by NCI P30 CA138313 (Hollings Cancer Center). Animal projects were conducted in a facility constructed with support from NCRR C06 RR015455. For their technical assistance, the authors give thanks to Lori Turner, David Holman, Sherry Zhou, William Meacham, Marisa Meyers-Needham, Cynthia Schandl, Laura Colombo, Kiki Gibbs, Ken Vanek, and Liya Zhou. The authors are indebted to Christina Voelkel-Johnson and Tejas Tirodkar for the AdCerS6-GFP vector and anti-CerS6 antibody; Marion Sewer for pGL3-pASAH1-Luc constructs; Michael Birrer for the pcDNA3-TAM67 construct; Jacek Bielawski for LC-MS/MS lipid analyses; Angen Liu for blinded immunohistochemistry scoring; and Yusuf Hannun, Dennis Watson, Keith Kirkwood, Besim Ogretmen, Elizabeth Garrett-Mayer, Xiaoyi Zhang, and Jennifer Wu for critical input.
Conflict of interest: Alicja Bielawska and James S. Norris hold stock in SphingoGene Inc., which is trying to develop LCL521 as a clinical drug for treating cancer.
Citation for this article: J Clin Invest. 2013;123(10):4344–4358. doi:10.1172/JCI64791.