Cancer Cell Biology
Cytotoxic action of phorbol esters on human pancreatic cancer cells
Version of Record online: 20 JUN 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 121, Issue 7, pages 1445–1454, 1 October 2007
How to Cite
Bond, J. A., Gescher, A. J., Verschoyle, R. D., Lemoine, N. R., Errington, R., Wiltshire, M., Smith, P. J. and Wynford-Thomas, D. (2007), Cytotoxic action of phorbol esters on human pancreatic cancer cells. Int. J. Cancer, 121: 1445–1454. doi: 10.1002/ijc.22869
- Issue online: 24 JUL 2007
- Version of Record online: 20 JUN 2007
- Manuscript Accepted: 4 APR 2007
- Manuscript Received: 16 NOV 2006
- Cancer Research UK
- Medical Research Council
- cell death;
- phorbol ester
We previously showed that phorbol esters are cytotoxic to human thyroid epithelial cells expressing a mutant RAS oncogene. Here we explore the generality of this finding using cells derived from pancreatic cancer, which, like thyroid, shows a high frequency of RAS mutation, but is a much greater cause of cancer mortality. The response to phorbol myristate acetate (PMA) and related agents was assessed on a panel of 9 pancreatic cancer cell lines, using a range of assays for cell growth and death in vitro and in vivo. In most lines, PMA induced non-apoptotic cell death which was, surprisingly, independent of its “classic” target, protein kinase C. With 24 hr exposure, the IC50 in the most sensitive line (Aspc-1) was <1 ng/ml (1.6 nM), with survival undetectable at concentrations ≥≥16 nM, and after only 1 hr exposure the IC50 was still ≤≤16 nM. Interestingly, the efficacy of a second phorbol ester, phorbol dibutyrate, was much lower, and the PMA analogue bryostatin-1, which is in clinical trials against other tumour types, was totally inactive. Pre-treatment of Aspc-1 cells with PMA before subcutaneous inoculation into nude mice prevented, or greatly retarded, subsequent xenograft tumour growth. Furthermore, treatment of established tumours with a single peri-tumoral injection of PMA induced extensive cell death and arrested tumour development. Taken together with recent Phase 1 clinical studies, these data suggest that activity against pancreatic cancer will be attainable by systemic administration of PMA, and point to potential novel therapeutic targets for this highly aggressive cancer. © 2007 Wiley-Liss, Inc.
Activating point mutations of the RAS oncogene family occur at high frequency in several types of human epithelial tumour, notably those of colon,1 pancreas2, 3 and thyroid.4 Indeed, in the latter 2 tumour types, this is probably a frequent initiating molecular event,5, 6 a hypothesis supported by our finding that experimental introduction of a mutant RAS gene into normal human thyroid epithelial cells induces an in vitro phenotype consistent with the corresponding early stage tumour (adenoma) in vivo.7, 8
As part of our initial exploitation of this model of epithelial tumour initiation, we had investigated the effect of phorbol esters on RAS-induced thyroid epithelial proliferation, anticipating a synergistic action on the basis of their known tumour promoting activity in keratinocytes.9 Surprisingly, however, both phorbol-myristate-13-acetate (PMA) and phorbol dibutyrate (PdBU) failed to stimulate growth, and instead killed cells expressing mutant RAS, whilst having no effect on normal cells.10 The possibility that this differential toxicity simply reflected the much higher proliferative rate of primary cells expressing mutant H-RAS was excluded using thyroid cell lines containing an inducible mutant RAS gene, treatment of which with PMA resulted in cell death only when RAS expression was induced, and independent of cell proliferation rate.11
We subsequently showed12 that the cytotoxic effect of PMA could be blocked by bisindolylmaleimide (GF109203X), a well-characterised inhibitor of “conventional” and “novel” (c and n) PKC isoforms, and by prior downregulation of PKC. This indicated that it is mediated by acute stimulation of PKC, as distinct from the subsequent downregulation13 thought to be responsible for the tumour promoting activity of long-term phorbol ester exposure.14 Western analysis identified two candidate PKC isoforms – α and ε.
Given the obvious therapeutic potential for tumours expressing mutant RAS, we have now proceeded to test the generality of this phenomenon using cell lines derived from pancreatic cancer, which has one of the highest frequencies of RAS mutation3 and is of course a much greater clinical problem than thyroid cancer, both in terms of incidence and mortality rate.15
Material and methods
Chemicals and reagents
PMA and PdBU (Sigma, Poole, UK), bryostatin (kindly donated by Dr. McGown, Paterson Institute, Manchester, UK) and bisoindolylmaleimide GF109203X (CN Biosciences, Beeston, UK) were dissolved in dimethylsulphoxide (DMSO) (Sigma) at stock concentrations of 1.6 mM and stored at −20°C.
Cells and culture conditions
The following human pancreatic cancer cell lines were used: Bxpc-3, Panc-1, MiaPaCa2, Aspc-1, HS766T (all from American Type Culture Collection); 818.4 and PT45 (gift from Dr. Holger Kalthoff (University of Kiel, Germany); T3M4 (gift from Dr. T. Okabe, Tokyo) and HPAF (gift from Dr R.S. Metzgar, Mainz, Germany). All cells were grown in RPMI1640 medium (Autogen BioClear, Calne, UK) supplemented with 10% fetal calf serum (Imperial Labs, London, UK).
Colony survival assays
Forty-eight hours after plating (2 × 105 per 60 mm dish) cells were re-fed medium containing PMA (and/or other agents). One to 24 hr later, monolayers were washed, trypsinised and replated at “clonal” densities (103, 5 × 103, 104 and 5 × 104 per 100 mm dish to allow for variation in plating efficiency). Ten to 21 days later (depending on rate of colony growth), dishes were fixed in methanol/acetic acid, stained with Giemsa and colonies (>50 cells) counted. Results were expressed as percentage of colony number in the corresponding controls exposed to the vehicle (DMSO) alone.
Cell counts and serial observations
Cells were seeded in replicates at 2 × 104 per cm2 (usually in 6-well plates) and 48 hr later exposed for the desired length of time to medium containing the appropriate concentration of PMA. Cultures were then washed, re-fed normal medium, observed daily by phase contrast microscopy and replicates trypsinised for cell counting at intervals up to 7 days.
To test for propidium iodide (PI) exclusion, cells were incubated in PI at 20 μg/ml in Hanks balanced salt solution (HBSS) for 5 min at room temperature, washed and examined either by conventional fluorescence or confocal (Bio-Rad 1024-MP) microscopy. Cells were also examined by confocal microscopy after incubation for 15 min in 5 μM Calcein-AM (Molecular Probes, OR), a fluorogenic esterase substrate which is hydrolysed to a green fluorescing product (calcein) that is retained in cells with an intact plasma membrane.16 Cells were considered non-viable if PI-positive or calcein-negative.
To assess nuclear fragmentation, cells were fixed in 4% paraformaldehyde for 10 min then washed, incubated in 1 μg/ml Hoechst 33258 (Sigma) for 5 min and examined by fluorescence microscopy. Alternatively, unfixed cells were incubated for 5 min in 5 μM DRAQ5 (Biostatus, Leicester, UK), a recently introduced cell-permeantfluorescent DNA binding probe.17
Terminal deoxynucleotidyl transferase (TdT) assay for DNA breaks was performed as described previously.18 Caspase-3 activation was assessed by immunocytochemistry using polyclonal rabbit antibody AF835 (R&D Systems, Abingdon, UK) specific for the activated form of the enzyme, together with a peroxidase detection system for light microscopy as described.19
Cells were treated with ribonuclease, stained with ethidium bromide and DNA content analysed on a FACS Vantage flow cytometer, as described previously.18
Analysis of MARCKS phosphorylation by Western blotting
Cells were lysed for 5 min on ice by 1% NP40 in 150 mM NaCl, 50 nM Tris (pH 8.0) and 5 mM EDTA buffer, which contained 1 mM phenylmethylsulfonyl fluoride, 50 mM sodium fluoride, 0.2 mM sodium orthovanadate, 20 μg/ml aprotinin and 10 μg/ml leupeptin and lysates were then cleared by centrifugation at 20,000g for 30 min at 4°C. Protein samples (20 μg) were separated by sodium dodecyl sulphate polyacrylamide gel electrophoresis on 8 or 10% gels, and electroblotted to Transblot polyvinylidene difluoride membrane (Bio-Rad Labs, Hemel Hempstead, Herts, U,K). Anti- phospho-MARCKS or MARCKS polyclonal antibodies (from Cell Signalling Technology, Beverley, MA and Santa Cruz Biotechnology, Santa Cruz, CA, respectively) were applied to the membranes at dilutions of 1/100. These were followed by anti-rabbit (for phospho-MARCKS) or anti-goat (for MARCKS) peroxidase conjugate at a dilution of 1/2,000 and were visualised by the ECL detection system (Amersham, Little Chalfont, UK). The membranes were stained with India ink to assess protein loading.
Aspc-1 or Panc-1 cells (suspended in 0.1 ml of HBSS) were injected subcutaneously into the flank of male ICRF nude mice (2 × 106 cells per site; one site per animal). Tumour volume was estimated as V = [L × W] × [(L + W)/2], where L and W are length and width, respectively, as assessed by calliper measurement. After termination of the experiment, tumours were excised, weighed, fixed in formalin and processed for conventional histological examination.
For in vivo treatment, PMA (50 μg/kg) was administered in 0.1 ml of 1% DMSO by a single s.c. injection immediately adjacent to one marked point on the tumour circumference. Controls received the same volume of 1% DMSO alone. Mice were examined daily and regular assessments of tumour volume were carried out to ensure compliance with the UKCCCR code of practice, and in accordance with the stipulations of Project Licence 80/1250 issued by UK Home Office.
Cytotoxic action of PMA on pancreatic cancer cell lines
We chose a series of 9 well-characterised spontaneously immortal cell lines derived from human pancreatic adenocarcinomas: Bxpc-3, Panc-1, MiaPaCa2, 818.4, PT45, T3M4, HS766T, HPAF and Aspc-1.20, 21 To permit comparison with our previous thyroid studies,12 we initially used the same “treatment” protocol which consisted of adding PMA to a final concentration of 1 μg/ml (1.6 μM) at day 0 and renewing this at subsequent re-feeding on day 3. Observation by phase-contrast microscopy (Fig. 1a) showed a range of responsiveness, massive cell loss ensuing within a few days in some lines whereas in others there was little or no visible effect. A similar range of sensitivity was seen when (for reasons given below) cells were exposed to PMA (1.6 μM) for just 1 hr, and cell numbers counted 7 days later (Fig. 1b). Interestingly, there was no apparent correlation with Ras gene mutation status (Fig. 1b).
To check for possible bell-shaped dose-response characteristics, a wide range of PMA concentrations was tested on a panel of lines, using a standard “colony survival” assay. Again, a similar rank-order of sensitivity was observed, at all PMA concentrations (Fig. 1c). Importantly, no effect of PMA beyond a slight slowing of growth was observed in primary cultures of normal human diploid fibroblasts (data not shown).
Aspc-1 and Bxpc-3 were chosen as the extreme examples of “sensitive” and “resistant” lines for further study.
Pharmacodynamics of response to PMA (Fig. 2a)
At the outset of this work, no reliable data were available on the pharmacokinetics of phorbol esters in man and preliminary analysis of stability in plasma in vitro indicated a half-life of no more than 30 min (Donald and Gescher, unpublished data). Given the anticipated difficulty, therefore, in achieving sustained high concentrations of PMA by systemic administration, we investigated in detail the minimum concentration and duration of exposure required for maximum cytotoxic action against pancreatic cancer cells in vitro, using the Aspc-1 line in a standard “colony survival” assay. With an exposure time of 24 hr, it was evident that the previously used dose of 1 μg/ml (1.6 μM) was well above the maximum required. Indeed colony survival was undetectable with doses ranging from this value down to as little as 10 ng/ml (16 nM). The IC50 for reduction in colony survival by PMA was estimated as less than 1 ng/ml (1.6 nM), the lowest concentration tested. When the exposure time was reduced, there was very little loss of efficacy with the highest concentrations (100 and 1,000 ng/ml) and even at the shortest exposure tested, 1 hr, the IC50 was still no more than 10 ng/ml (16 nM). For subsequent experiments, 1 μg/ml for 1 hr was chosen as a convenient maximally effective “standard” protocol. No reduction in colony survival was seen at any dose/time combination in the Bxpc-3 cell line.
Lower efficacy of PdBU (Fig. 2b)
A second phorbol ester, PdBU was examined because in most published work, including our previous studies on thyroid cells,11 it exhibits very similar activity to PMA, and its much lower lipophilicity22 might be expected to confer more advantageous pharmacokinetic properties in therapeutic use. Surprisingly, however, in the context of pancreatic cancer cell cytotoxicity PdBU was much less effective than PMA. Even with 24 hr exposure, the colony survival of Aspc-1 was never less than 25% of the DMSO control, compared to 0% for PMA. At shorter times the difference was even more marked, 1 μg/ml being required to achieve just a 50% reduction in colony yield. Again, no effect was observed on the Bxpc-3 cell line.
Inactivity of bryostatin-1 (Fig. 2c)
Many of the biological effects of phorbol esters can be mimicked by the macrocyclic lactone, bryostatin-1.23 Since this agent lacks tumour-promoting properties in the experimental mouse skin model and has already reached Phase II clinical trials for several tumour types,24, 25 we investigated its effectiveness against pancreatic cancer cells. To maintain comparability with the PMA results, the same range of molarity was used, that is, 1.6, 16, 160 and 1,600 nM (corresponding to 1–1,000 ng/ml PMA), which spans the range in which its biological activity against cancer cells has been previously reported. No effect on survival of Aspc-1 (or other pancreatic cancer lines) could be detected at any concentration after either 1, 6 or 24 hr exposure. To exclude trivial causes of this negative result, we confirmed the activity of bryostatin-1 in our hands by showing the expected cytostatic effect on MCF-7 breast cancer cells, as revealed by a marked reduction in cell counts 7 days after a 24 hr exposure to bryostatin at either 10 or 100 nM (data not shown), consistent with previous reports.26
Nature of cytotoxic action of PMA on pancreatic cancer cells
Phase contrast microscopy (Fig. 1a) and cell counts (Fig. 1b) clearly indicated that the effect of PMA on Aspc-1 cultures was due largely to cell loss rather than just growth arrest. This was confirmed by more detailed sequential observations (Fig. 3a) which showed that cells began to round up within a few hours after PMA treatment (1 μg/ml; 1 hr) and by 48 hr most were found in loosely-attached clumps. Confocal microscopy showed these to be arch-like structures anchored to the dish by just a few of their component cells, which subsequently detached completely over the following few days. Further examination of these clumps at 48 hr (Fig. 3b) showed that the participating cells were nearly all viable, as evidenced by cytoplasmic fluorescence after staining with calcein-AM (Fig. 3b) and absence of uptake of PI (not shown). Furthermore, they also appear to be still proliferating as assessed both by flow cytometric analysis of DNA content (Fig. 3b) and BrdU incorporation (nuclear labelling index: 11–13% in PMA-treated compared to 35–40% in untreated controls). There was also very little evidence of apoptosis at any time point following PMA treatment, as assessed by normal nuclear morphology following staining with the DNA binding fluorochrome DRAQ5 (Fig. 3b), by absence of a sub-G1 peak on flow cytometry (Fig. 3b), and by TdT assay or immunostaining with antibody to activated caspase-3,19 less than 1% cells being positive in both treated and control cultures by both the latter assays (data not shown). In contrast, examination of the detached cells recovered from the tissue culture medium 5 days post-PMA treatment showed that the majority labelled with PI (Fig. 3c) and less than 1% incorporated BrdU (not shown). No more than 5% showed evidence of apoptosis either by caspase-3 immunostaining (not shown) or nuclear morphology (Fig. 3c). Pre-treatment with the caspase inhibitor z-VAD-fmk (Alexis Biochemicals) (20 or 100 μM added 4 hr before PMA) failed to exert any effect on the subsequent response of Aspc-1 cells to PMA (1 μg/ml; 1 hr) as assessed by phase contrast microscopy or final cell counts (data not shown).
Taken together, these data indicate a chain of events in which 1hr exposure to PMA triggers over the first few days a progressive rounding up of cells which initially remain viable and cycling, but which subsequently detach and at some point thereafter undergo a predominantly non-apoptotic form of cell death.
Evidence for a PKC-independent mechanism of action of PMA
To obtain an initial insight into the relevant signal pathway(s), we examined the effect of the bisindolylmaleimide GF109203X, an inhibitor of c and n isoforms of PKC which we had previously found to efficiently block the cytotoxic action of PMA on thyroid cells expressing mutant RAS.12 GF109203X (at our previously used concentration of 1.6 μM or the higher concentration of 5 μM used in many other studies) was added to Aspc-1 cells 1 hr before, or at the same time as, treatment with PMA (1.6 μM for 1 hr). Cells were re-seeded at 5 × 104, 104 or 5 × 103 per 100 mm dish and the subsequent yield of colonies assessed 10–21 days later. As shown in Table I, no protective effect of GF109203X could be detected in any experimental condition, although interestingly there was a slight enhancement of colony yield (and size) following exposure to the inhibitor alone (without PMA treatment).
|Cells plated||Mean colony yield (per dish)|
|5 × 104||+++++2||0||0||0||0||0||++++++||+++++|
|5 × 103||53||0||0||0||0||0||133||23|
To confirm the effectiveness of GF109203X as a PKC inhibitor in these experiments, we analysed its effect on phosphorylation of a well-characterised substrate—MARCKS27—as a “read-out” of PKC activity in the intact cell. Western blots using antibodies against phospho-MARCKS or total MARCKS protein (Fig. 4) showed that GF109203X at 1.6 μM prevented the stimulation of phosphorylation by PMA, and at 5 μM completely abolished even basal levels of activity.
We conclude therefore that the cytotoxic effect of PMA was not dependent on stimulation of PKC. This is also consistent with dose-response data obtained using the MARCKS assay (not shown). This showed that phosphorylation following 1 hr exposure to PMA was of equally high magnitude and duration over a wide range of concentrations—from 1.62 nM to 1.62 μM (1 μg/ml)—in sharp contrast to the marked concentration-dependence seen for the effect of PMA on colony survival in Figure 2a.
An alternative possibility13, 14 is that PMA exerts its effect, not through stimulation but via subsequent downregulation of PKC isoforms. We considered this highly unlikely since if this were the mechanism, we would have expected pharmacological inhibition of PKC by GF109203X to also have induced cytoxicity, which as shown in Table I was not the case. However, to provide further evidence, we carried out several detailed time-course studies of PKC activity following a cytotoxic exposure to PMA (1 μg/ml for 1 h), one of which is shown in Figure 5. Maximum response, as reflected by MARCKS phosphorylation, is reached by 1 h, is sustained for a further 3–6 hr and then returns to baseline between 24 and 48 hr. At no time up to 24 hr did we observe a fall to below basal levels.
Although MARCKS phosphorylation status is an indirect measure of PKC activity, the speed of response to PMA noted above and the subsequent rate of turnover of phosphorylation of MARCKS following addition of GF190203X (which indicated a half-life of less than 30 min—data not shown) strongly suggest that changes in the level of phospho-MARCKS can be taken to reliably reflect changes in PKC activity, with a lag of no more than a few hours. Since the cytotoxic effect of PMA is already well established by 12 hr after exposure, and since the level of MARCKS phosphorylation is still well above baseline at 24 hr (Fig. 5), we can confidently exclude downregulation of PKC as a contributory mechanism.
Inhibition by PMA of xenograft tumour growth in nude mice (Fig. 6)
To avoid the inherent uncertainties of in vivo administration, we initially chose a sequential in vitro/in vivo approach to assess the ability of PMA to block growth of pancreatic cancer cells as a xenograft. Aspc-1 cells were pre-treated for 1 hr with either 1 μg/ml PMA or DMSO alone (control), washed, trypsinised and then inoculated subcutaneously into nude mice (2 × 106 cells per site; 1 site per animal). All 10 controls generated rapidly growing tumours (Fig. 6a). One was killed at 3 weeks due to early skin invasion, the other 9 when maximum permissible diameter began to be approached in the most rapidly growing cases (Day 38). At this stage, tumour weights ranged from 113 to 435 mg with a mean of 327 ± 45 mg (SEM). Pre-treatment of Aspc-1 cells with PMA dramatically delayed tumour growth in all 9 animals inoculated (Fig. 6b). At 3 weeks, only one had a clearly detectable tumour. One died with no visible tumour at 3 weeks (cause unknown). When the experiment was terminated (at day 38), of the remaining 8 mice, 1 still had no visible tumour, 4 had only very small nodules (<< 50 mg) and only in the other 3 cases had tumour growth “caught up” to reach sizes within the control range (99.8, 139 and 468 mg, respectively).
Similar results were obtained using a second cell line, Panc-1 (Fig. 6c). Although less tumorigenic than Aspc-1 (as revealed by the overall slower and more variable tumour growth), pre-treatment with PMA again had a major inhibitory effect on tumour development. Whereas 5 out of 10 controls reached volumes between 300 and 1,100 mm3 when the experiment was terminate (Day 65), only one pre-treated case exceeded 200 mm3 and 5 out of 11 were invisible or less than 50 mm3. The difference in final average tumour sizes was highly significant: 414.9 ± 109 vs. 52.7 ± 21.5 (mean ± SE) in control and treated groups respectively (p < 0.001).
Activity of PMA against established xenograft tumours (Fig. 7)
Since it was formally possible from the above that PMA merely prevented xenograft “take”, activity against established tumours was investigated directly. To avoid the practical difficulty and potential variability of the i.v. route in mice, PMA was administered by a single peri-tumoral s.c. injection immediately adjacent to the tumour.
In the first approach (Fig. 7a) Aspc-1 cells were inoculated s.c. as above, and tumours allowed to develop for 18 days (to an average size of 98 mm3) before peri-tumoural injection of PMA (50 μg/kg) or vehicle only (control). Subsequent serial caliper measurements showed that in contrast to the control group, which continued to grow until the termination of the experiment, tumours receiving a single dose of PMA underwent an arrest of growth which was maintained for at least 9 days followed by a gradual “catch up”.
To provide a direct assessment of the acute response to PMA, a second series of tumours were allowed to grow for 40 days before receiving PMA or vehicle only, and then removed and processed for routine histology 4 days later. Again, there was an indication of tumour growth inhibition, though not surprisingly given the very short follow-up period, this difference was not statistically significant, the mean volume at Day 44 being (99.8 ± 8.1)% of the pretreatment (Day 40) value in the PMA group (n = 12) compared to (115 ± 5.7)% in controls (n = 10). More importantly, histological examination of tissue sections (Fig. 7b) showed that in addition to the background of focal ischemic necrosis seen in controls (and in unmanipulated tumours), many PMA-treated tumours showed large confluent areas of recent cell death. These were most marked in the 5 of the 12 TPA treated tumours which showed greater than average reductions in tumour volume after 4 days (ranging from 60 to 80% of the pretreatment value). Importantly, there was no evidence of cell death in the surrounding non-neoplastic tissues.
Although initially considered to have predominantly anti-apoptotic actions,28 phorbol esters were, in contrast, shown by our laboratory to be cytotoxic for human thyroid epithelial cells expressing mutant RAS, pointing to their potential as anti-cancer agents.10, 11 Since then, similar responses have been reported in several tumour cell types, including prostate29, 30 and gastric.31 Our results together with similar work published independently by Avila et al.32 add pancreas to this growing list of potential tumour targets.
Interestingly, there are significant differences in the mode of cell death induced by PMA in these different cell types. Whereas LNCAP prostate cancer cells displayed a classic rapid apoptotic response,29, 33 in thyroid cells expressing mutant RAS, death was more protracted and lacked some of the hallmarks of apoptosis.12 Our present data indicate that in pancreatic cancer cells the primary effect appears to be disruption of cell-substratum adhesion, without immediate loss of viability; subsequently cells detach and finally die with characteristics consistent with necrosis rather than apoptosis (resembling the response of LNCAP cells treated with ceramide in place of PMA34).
If this were merely a tissue culture phenomenon, it would be of limited therapeutic interest. However, our in vivo data show that even a brief pre-treatment with PMA in vitro immediately before inoculation is sufficient to prevent subsequent growth of pancreatic cancer cells as a xenograft. In other words, PMA irreversibly engages a cell death response which continues to be executed even in an in vivo tissue environment. Moreover, we have shown direct activity against established tumours from just a single in vivo administration of PMA, as revealed by histological evidence of widespread cell death and by prolonged cessation of overall tumour growth. These results strongly support the findings of Avila et al.32 who also demonstrated a dose-dependent toxic action of PMA on the Panc-1 pancreatic cancer cell line in culture (though interestingly a partial response was also seen in Bxpc-3 cells). Furthermore, these authors also found that daily i.p. administration of PMA strongly inhibited growth and/or induced regression of Panc-1 (but not Bxpc-3) tumour xenografts in vivo, associated with an increase in apoptosis. Similar anti-tumour effects of PMA were previously reported by this group on prostate cancer xenografts.30
In thyroid12 and LNCAP cells,34 use of inhibitors such as GF109203X has shown that the cytotoxic action of PMA is mediated by one or more isoforms of c- or n-PKCs. Also in one previous study of PMA effects on pancreatic cancer cells,35 PKC-dependence was observed, although this was only assessed with respect to a short-term, partial cell-cycle inhibitory response to PMA and in a different cell line (DanG). Unexpectedly, we have shown here, using longer-term assays, that PMA exerts a cytotoxic effect on pancreatic cancer cells which is not inhibited by GF109203X, even at concentrations sufficient to abolish detectable PKC activity, strongly suggesting that activation of PKC is not required. Furthermore, we found that the onset of cytotoxicity in response to a 1 hr exposure to PMA occurs while PKC activity is still well above baseline, and that GF109203X-mediated inhibition of PKC does not result in any cell death, both findings arguing against downregulation as the mechanism of action. Taken together therefore we conclude that PMA is acting here through a PKC-independent pathway.
Recently, several targets have been identified which share the C1 phorbol ester binding-site of PKC, but lack its kinase domain.36, 37 Ras GRP38 is an attractive candidate, since it may mimic the effect of PKC on RAS signalling. However, the available data38 suggest its expression is limited to brain and haemopoietic cells, and not epithelia, although pancreas appears not to have been specifically studied; we also failed to detect expression of the protein by immunocytochemistry in Aspc-1 cells (data not shown). Beta 2-chimerin is a Rac-GAP which again may interact with RAS signalling.39 PKD1 is involved in intracellular membrane trafficking and has also been implicated in cell proliferation and invasion.40 Munc13 is another potential candidate which, intriguingly, was shown to confer phorbol ester induced apoptosis when artificially overexpressed in renal epithelium.41 Further work will be needed to explore which, if any, of these known candidates might be responsible for the apparent change in cell adhesion which precedes cell death in this system. Confusingly, phorbol esters have previously been reported to increase adhesiveness in haemopoietic and some epithelial cell lines by stimulating integrin expression/function,42, 43 but such paradoxes are typical of the diversity of biological responses to these agents.28
Unexpectedly, the wide variation in initial responsiveness observed between different pancreatic cancer lines could not be accounted for by the presence or absence of RAS mutation, which contrasts with our earlier finding in thyroid cells that mutant RAS conferred sensitivity to the cytotoxic effects of PMA. However, the significance of this finding is difficult to assess. Given that the majority of clinical pancreatic cancer samples have a K-RAS mutation the finding that a few wt lines (such as T3M4) are highly sensitive to PMA may of course be explicable by their having acquired an alternative oncogenic mutation, functionally equivalent to mutant RAS. Conversely, the low sensitivity of 2 out of the 6 mutant lines cannot readily be ascribed to RAS status and an explanation will need to await identification of the relevant PMA target.
Nevertheless, assuming that at least some naturally occurring pancreatic cancers will behave similarly to the 4 out of 6 mutant RAS lines (such as Aspc-1) that were highly sensitive to PMA, the pharmacodynamic data are encouraging with respect to therapeutic use. Thus we have shown that PMA produces near-maximal cytotoxic response at a concentration (1–10 ng/ml) similar to that reported for many other biological effects of phorbol esters, but with a much shorter minimum exposure time than had been anticipated from our previous thyroid studies. This makes it much more likely that the in vivo delivery (area under the plasma concentration vs. time curve; AUC) required for anti-tumour activity will be attainable. Indeed, there is support for this from Phase 1 studies of PMA administration to leukaemia patients (see below).44
Surprisingly, and in contrast to previous findings in thyroid,11 PdBU was much less effective than PMA. At shorter exposure times, this might be partly due to differential intracellular retention following removal from the medium, a corollary of the difference in hydrophobicity between PMA and PdBU.22 However this is unlikely to explain the inferior efficacy of PdBU even with protracted (24 hr) exposure. The macrocyclic lactone, bryostatin-1, was also tested since it mimics many actions of phorbol esters (both via PKC and non-PKC targets) and has undergone clinical evaluation against several common cancers, both haematological and epithelial.24, 25 It would therefore have been a preferable agent since much more is known of its potential toxicity in man, and it has the additional (although we would argue minor) advantage of lacking tumour-promoting activity in mouse skin models. Unfortunately, however, we have shown conclusively that bryostatin-1 is devoid of activity against pancreatic cancer cells in our assays.
Although all three agents tested here have been shown to bind with similar affinity to most PKC and non-PKC targets so far identified,36 widely divergent biological responses are not uncommon and recently have been attributed to differential intracellular translocation. For example, PKC delta can be translocated from cytoplasm to plasma membrane, nucleus or even Golgi depending on the nature of the fatty acid side chains to which the phorbol is esterified.22 Again, a test of this explanation for our results will need to await identification of the relevant PMA target but may have important implications for eventual design of PMA analogues for clinical use.
Probably because of their historical use as tumour promoters, phorbol esters have been overlooked as anti-cancer agents in man. This is unfortunate, since the promoting effect requires repeated treatment over many weeks, and probably reflects a different mechanism from the transient exposure required to elicit the cytotoxic responses described here. Indeed, any such risk must surely be outweighed by the dismal prognosis faced by pancreatic cancer patients.15
Although still few in number, several clinical studies on PMA have now been published.44, 45, 46, 47, 48 The first, performed in China, involved multiple i.v. administrations of PMA (up to 1.0 mg over 1 hr) to patients with acute myeloid leukaemia45, 46 and although no pharmacokinetic analysis was performed, biologically active systemic concentrations appear to have been achieved, since both therapeutic and toxic effects were observed. A subsequent Phase 1 trial in the USA also revealed an acceptable toxicity profile in leukaemia patients.44 Most importantly, direct measurement in later trials using a bio-assay,44, 47, 48 showed that peak blood concentrations following i.v. infusion in man were comparable to those which were associated with successful inhibition of pancreatic32 and prostate30 cancer xenograft growth in mice. In addition, the peak levels and half-life of PMA achievable in man44, 47, 48 were such that the concentration-time profile in many of these patients should have been well within the range required to induce growth inhibition and/or cell death in pancreatic cancer cells in culture (this study and32).
Taking these emerging clinical data together with the experimental evidence presented here and in Avila et al.,32 and given the persistent refractoriness of pancreatic cancer to existing agents, we suggest that further investigation of the molecular target(s) and therapeutic potential of PMA in pancreatic cancer is now justified.
We are grateful to Cancer Research UK and the Medical Research Council for grant support, Lynn Howells for technical assistance and Julia Parmar for manuscript preparation.
- 5The molecular genetics of pancreatic adenocarcinoma. In: NeoptolemosJP,LemoineNR. eds. Pancreatic cancer. Oxford: Blackwell Scientific Publications, 1996; 115–32., .
- 17Characteristics of a novel deep red/infrared fluorescent cell-permeant DNA probe, DRAQ5, in intact human cells analyzed by flow cytometry, confocal and multiphoton microscopy. Cytometry 2000; 40: 280–91., , , , , , , , , .
- 32Inhibitory effects of 12-O-tetradecanoylphorbol-13-acetate alone or in combination with all-trans retinoic acid on the growth of cultured human pancreas cancer cells and pancreas tumor xenografts in immunodeficient mice. J Pharmacol Exp Ther 2005; 315: 170–87., , , , , , , , , , , , et al.