Statins are lipid-lowering drugs that may help limit cancer occurrence in humans. They drive blockage of the mevalonate pathway, trigger cancer cell apoptosis in vitro and reduce tumour incidence in animals. We have shown in the present study that statins induced apoptosis in HGT-1 human gastric cancer cells, and this was prevented by intermediates of the cholesterol synthetic pathway. In addition, similarly to what we have reported previously for caspase 2 [Logette, Le Jossic-Corcos, Masson, Solier, Sequeira-Legrand, Dugail, Lemaire-Ewing, Desoche, Solary and Corcos (2005) Mol. Cell. Biol. 25, 9621–9631], caspase 7 may also be induced by statins and is under the positive control of SREBP (sterol-regulatory-element-binding protein)-1 and -2, major activators of cholesterol and fatty acid synthesis genes, in HGT-1 cells. Knocking down these proteins strongly reduced caspase 7 mRNA and protein expression, and chromatin immunoprecipitation analyses showed that the proximal promoter region of the CASP7 gene could bind either SREBP-1 or -2. Strikingly, cells selected to grow in the continuous presence of statins showed increased expression of caspase 7 mRNA and protein, which was maintained in the absence of statins for several weeks, suggesting that high expression of this caspase might participate in adaptation to blunting of the mevalonate pathway in this model. Taken together, our results show that caspase 7, as an SREBP-1/2 target, can be induced under mevalonate-restricting conditions, which might help overcome its shortage.
Statins are HMG-CoA (3-hydroxy-3-methylglutaryl-CoA) reductase competitive inhibitors that have long been used as efficient lipid-lowering drugs in humans, but a large body of accumulated evidence suggests that they bear potential as anticancer agents in adjuvant therapy protocols [1–4]. Statins have been shown to trigger cancer cell apoptosis, and to decrease tumour incidence, tumour size or metastasis potential in animal models [3,5,6]. This was associated with a reduction in expression of the anti-apoptotic Bcl-2 and survivin proteins, up-regulation of pro-apoptotic members of the Bcl-2 family and caspase activation [7–9]. From a mechanistic point of view, statin-dependent apoptosis has been shown to be prevented by the addition of mevalonate, the product of HMG-CoA reductase activity .
Statin treatment leads to suppression of cholesterol and fatty acid synthesis, which triggers a compensatory feedback mechanism that aims at restoring lipid synthesis upon induction of many, if not all, genes encoding lipid biosynthetic enzymes and the LDLR [LDL (low-density lipoprotein) receptor] [11,12]. This feedback mechanism is orchestrated by SREBPs (sterolregulatory-element-binding proteins), which become transcription activators when membrane cholesterol depletion is sensed, following transfer of their processed N-terminal fragment into the nucleus [13,14]. This results in stimulation of lipid anabolism and uptake, and helps cells to restore their membrane lipid pool. Two genes encode SREBPs. The SREBF-1 gene encodes SREBP-1a and SREBP-1c, which differs from the former by a shorter first exon that codes for the N-terminal trans-activating function of the mature protein . SREBP-2 is encoded by a different gene (SREBF-2) which codes for only one isoform . The activities of SREBP-1 and SREBP-2 partially overlap, but SREBP-1 is more involved in the control of fatty acid synthesis, whereas SREBP-2 is primarily involved in cholesterol synthesis in mice . From results in transgenic mice overexpressing SREBP-1 or -2, or deficient in sterol cleavage-activating protein, it was shown that most SREBP-responsive genes code for lipid biosynthetic enzymes, but also for some other genes which are apparently unrelated .
We previously demonstrated that the human CASP2 gene, encoding pro-caspase 2, was both an SREBP-responsive gene and an important determinant of lipid synthesis in several cancer cell lines [17,18]. In addition, the CASP2 gene was induced upon treatment with lovastatin, confirming that it was a bona fide SREBP-responsive gene in human cells.
In the present study, we looked at the potential responsiveness of other caspases to statins in the human gastric cancer cell line HGT-1 . Our results show that statins induced apoptosis in these cells, which was suppressed by intermediates of the mevalonate pathway and by both broad spectrum and specific caspase peptide inhibitors. We also identify CASP7 as a novel statin-responsive gene. ChIP (chromatin immunoprecipitation) experiments showed that SREBP-1 and -2 could bind to the proximal promoter region of the CASP7 gene, whose expression was suppressed upon silencing the SREBF-1 and -2 genes, indicating that the effect of SREBPs on the CASP7 gene might be essentially direct. In addition, adaptation of HGT-1 cells to growth in the presence of statins led to a stable rise in pro-caspase 7, suggesting that this caspase may help to overcome the blockage of the mevalonate pathway brought about by statin treatment.
MATERIALS AND METHODS
HGT-1 human gastric cancer cells were grown at 37 °C under a humidified atmosphere of 5% CO2 in DMEM (Dulbecco's modified Eagle's medium; Lonza), containing 4.5 g/l glucose and supplemented with 5% (v/v) foetal bovine serum without antibiotics (Gibco-Invitrogen) .
The human HCT116 colon carcinoma cells (A.T.C.C. number CCL-247) and the human HepG2 hepatoma cells (A.T.C.C. number HB-8065) were maintained in DMEM/Ham's F12 medium (Lonza) without antibiotics and supplemented with 5% (v/v) and 10% (v/v) foetal bovine serum respectively.
Selection of statin-resistant cell populations was performed with mass cultures grown in the presence of serum-containing medium and 25 or 50 μM lovastatin. Massive cell death occurred for several weeks under continuous selective pressure, after which the populations stabilized and started to grow with no more signs of apoptosis. Then, lovastatin was removed from the culture medium for several cell doublings and the cells were challenged again with 25 or 50 μM lovastatin. No more cell death occurred, and the population was considered to be stably resistant to lovastatin. This procedure was applied successfully to HGT-1 (50 μM lovastatin) and HCT116 cells (25 μM lovastatin).
All statins used in the present study were purchased from Calbiochem. They were dissolved in DMSO, except for fluvastatin and pravastatin which were dissolved in water. The final concentration of DMSO did not exceed 0.4%, a concentration that did not induce any toxicity. Camptothecin and mevalonate (Sigma–Aldrich) were dissolved in DMSO. Cholesterol (Sigma–Aldrich) was dissolved in ethanol. Caspase inhibitors were from R&D Systems and were dissolved in DMSO. FPP (farnesyl diphosphate) and GGPP (geranylgeranyl diphosphate) were purchased from Echelon Biosciences (Tebu-Bio) and were dissolved in water. The mevalonate concentration in cell culture medium or in pure foetal calf serum was determined according to Henneman et al. .
Determination of apoptotic fragmentation and caspase activation
The cells were treated with different doses of lovastatin or with camptothecin, which is known to induce apoptosis. Apoptosis was determined by Hoechst 33342 (10 μg/ml in PBS) staining of the cells for 15 min at 37 °C and fluorescence microscopy analysis of 300 cells per condition.
The activity of caspase 3 and caspase 7 was determined using the luminescent Caspase-Glo™ 3/7 Assay kit (Promega) according to the manufacturer's protocol. Briefly, the day before treatment, the cells were plated in 96-well plates (20000 cells per well). After treatment, an equal volume of Caspase-Glo™ 3/7 reagent was added to the sample in the assay well. Samples were incubated at 22 °C for 1 h and the enzyme activity was measured with a luminometer (Fluoroskan Ascent FL; Thermo Electron).
DNA fragmentation was analysed by separation of 3 μg of DNA in 1.8% agarose gels and staining with ethidium bromide. After treatment, cells were resuspended in lysis buffer [50 mM Tris/HCl (pH 8), 20 mM EDTA and 1% (w/v) SDS] containing 200 μg/ml proteinase K. After incubation at 56 °C for 16 h, DNA was purified by phenol/chloroform extraction and precipitated by the addition of 0.3 M sodium acetate and 2 vol. of ethanol. The precipitates were pelleted by centrifugation at 12000 g for 20 min at 4 °C, air-dried and resuspended in 10 mM Tris/HCl (pH 8) containing 1 mM EDTA.
RNA extraction and RT (reverse transcription)–PCR analysis
Total RNA was isolated using TRIzol® (Invitrogen). Samples of 2 μg of total RNA were reverse-transcribed using random hexamers as primers and MMLV (Moloney-murine-leukaemia virus) reverse transcriptase (New England Biolabs). RT reactions (1 μl out of a 25 μl reaction volume) were used as templates for all PCR experiments. The PCR primers used in the present study were: acidic ribosomal phosphoprotein (P0), 5′-GCGACCTGGAAGTCCAACTA-3′ (sense) and 5′-TCTCCAGAGCTGGGTTGTTT-3′ (antisense); HMG-CoA reductase (HMG-CoA red), 5′-CGATGCATAGCCATCCTGTA-3′ (sense) and 5′-TCAAGCCTGTCAATTCTTTGTC-3′ (antisense); FAS (fatty acid synthase), 5′-GCTGGGTGGAGTCTCTGAAG-3′ (sense) and 5′-TGCAACACCTTCTGCAGTTC-3′ (antisense); LDLR, 5′-GCTTGTCTGTCACCTGCAAA-3′ (sense) and 5′-AACTGCCGAGAGATGCACTT-3′ (antisense); FPPS (farnesyl pyrophosphate synthase), 5′-AGGGCAATGTGGATCTTGTC-3′ (sense) and 5′-GAAAGAACTCCCCCATCTCC-3′ (antisense); SREBP-1, 5′-TGACTTCCCTGGCCTATTTG-3′ (sense) and 5′-TTCAATGGAGTGGGTGCAG-3′ (antisense); SREBP-2, 5′-TGGACGTGGTGGTAGTGGTA-3′ (sense) and 5′-CTGAGGTGGGAGAAACCTTG-3′ (antisense); caspase 3 (Casp3), 5′-GACTCTAGACGGCATCCAGC-3′ (sense) and 5′-TGACAGCCAGTGAGACTTGG-3′ (antisense); caspase 6 (Casp6), 5′-GCCAGTCATTCCTTTGGATG-3′ (sense) and 5′-ATGAGCCGTTCACAGTTTCC-3′ (antisense); caspase 7 (Casp7), 5′-AGTGACAGGTATGGGCGTTC-3′ (sense) and 5′-CGGCATTTGTATGGTCCTCT-3′ (antisense); and caspase 9 (Casp9), 5′-GAGGGAGTCAGGCTCTTCCT-3′ (sense) and 5′-CTGGTCGAAGGTCCTCAAAC-3′ (antisense).
ChIP assays were performed using the ChIP-IT kit (Active Motif Europe) according to the manufacturer's protocol. Briefly, cells were fixed with formaldehyde for 10 min at 20 °C to cross-link DNA with proteins, and then sheared chromatin was prepared enzymatically with a 10 min incubation time. After incubation with Protein G magnetic beads, the supernatants were immunoprecipitated with 2 μg of 2A4 anti-SREBP-1 (Santa Cruz Biotechnology/Tebu) or N19 anti-SREBP-2 (Santa Cruz Biotechnology/Tebu) antibodies or with 2 μg of negative control IgGs at 4 °C overnight. The beads were then washed, the cross-links reversed and the samples treated with proteinase K. For PCR (33 cycles), 5 μl of eluted DNA or control input DNA was used. The primers used corresponded to the proximal promoter of the CASP7 gene and were as follows: OL1s, 5′-TTGGTCAGGGTGAACTGGAT-3′ (sense) and OL2as, 5′-GATGCTGCTCCTGGATCTTT-3′ (antisense). PCR on material recovered for immunoprecipitation without antibody was also performed as a control. PCR products were visualized on 2.5% agarose gels containing ethidium bromide.
Protein extraction and Western blot analysis
Treated cells were harvested, washed in PBS and lysed in boiling buffer [1% (w/v) SDS, 1 mM Na3VO4 and 10 mM Tris/HCl (pH 7.4)] containing protease inhibitor cocktail (Roche) for 10 min at 4 °C. Proteins (50 μg) were boiled in Laemmli buffer for 5 min, separated by SDS/PAGE (12% gels) and blotted on to PVDF membranes (Bio-Rad). Non-specific binding sites were blocked for 1 h at room temperature by 5% (w/v) fat-free skimmed milk powder before an overnight incubation at 4 °C with specific antibodies: rabbit anti-human pro-caspase 3, 6, 7 or 9 (Cell Signaling Technology), a rabbit anti-human PARP [poly(ADP-ribose) polymerase; Epitomics-Euromedex], mouse monoclonal anti-human SREBP-1a (2A4, Santa Cruz Biotechnology/Tebu), mouse monoclonal anti-human SREBP-2 (1C6; Santa Cruz Biotechnology/Tebu)or with a rabbit anti-human Hsc70 (heat-shock cognate 70 stress protein; Abcam) antibody as a loading control. Primary antibodies were detected with HRP (horseradish peroxidase)-conjugated anti-mouse, anti-goat or anti-rabbit IgGs (Amersham Biosciences). Blots were revealed using an ECL (enhanced chemilumiscescence) detection kit (Amersham Biosciences) by autoradiography.
siRNA (small interfering RNA) transfections
Sense and antisense SREBP-1 (sense strand RNA sequence, 5′-CAACCAAGACAGUGACUUC-3′), SREBP-2 (sense strand RNA sequence, 5′-CAACAGACGGUAAUGAUCAUU-3′), caspase 7 (sense strand RNA sequence, 5′-CCGUCCCUCUUCAGUAAGA-3′) and negative (ref: OR-0030-neg05) oligoribonucleotides were purchased from Eurogentec. HGT-1 cells were seeded into six-well plates at a density of 1×105 cells/well the day before transfection and then were transfected by adding 5 μl of Lipofectamine™ 2000 reagent to 5 μl of 20 μM siRNAs (final concentration 100 nM) in serum-free medium. At 4 h after transfection, the medium was supplemented with serum and then maintained in culture for 72 h before analysis.
RESULTS AND DISCUSSION
Statins trigger HGT-1 cell apoptosis
The cytotoxic activity of statins has been largely documented in several types of cancer cells; however, little is known about gastric cancer cells. In order to analyse the effect of statins in such cells, we used the HGT-1 cell line, which was isolated several years ago from an adenocarcinoma of the stomach . Several statins were used at concentrations of 25 and 50 μM. As observed with Hoescht 33342 staining, lovastatin, mevastatin, fluvastatin and simvastatin triggered nuclear fragmentation in 30–50% of cells, whereas pravastatin had only marginal activity (Figure 1A). The fact that pravastatin was relatively inefficient has already been observed in several other cell models . It was shown that apoptotic effects of pravastatin could be observed for very high concentrations (above 100 μM), which have not been used in the present study . In addition, the rise in caspase (3/7) activity, a biochemical marker of ongoing apoptosis, strictly followed the relative levels of cell death induced by the various statins, as determined by activation of caspase (3/7) cleavage-dependent luciferase (Caspase Glo™ reagent) (Figure 1B).
Apoptosis induction by statins in HGT-1 gastric cancer cells
To get a more complete picture of apoptosis induction in HGT-1 cells, we focused on the effects of lovastatin. Both dose- and time-dependency of apoptosis were observed upon lovastatin treatment: 15% apoptosis induction was detected with 5 μM lovastatin for 48 h, and 30% apoptosis was attained after 96 h with the same concentration of lovastatin, whereas more than 60% apoptosis occurred with 50 μM lovastatin for 96 h (Figure 1C). Inter-nucleosomal DNA fragmentation and PARP cleavage was also observed in response to lovastatin (Figures 2A and 2B). This apoptosis was strictly dependent on the activation of pro-caspases as the broad spectrum caspase inhibitor Z-VAD-fmk (benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone) fully suppressed cell death. Furthermore, Z-VDVAD-fmk (benzyloxycarbonyl-Val-DL-Asp-Val-Ala-DL-Asp-fluoromethylketone), a caspase 2 preferred inhibitor, reduced apoptosis by 86% and Z-DEVD-fmk (benzyloxycarbonyl-DL-Asp-Glu-Val-DL-Asp-fluoromethylketone), a caspase (3/7) inhibitor, reduced apoptosis by 93% (Figure 2C). Hence, several caspases accounted for the high susceptibility of HGT-1 cells to lovastatin-dependent apoptosis. When assayed by overexpression, we found that caspase 7 had only a limited contribution to apoptosis induction (results not shown), suggesting that the strong inhibitory effect of Z-DEVD-fmk on lovastatin-induced apoptosis could be better explained by an effect on active caspase 3 than caspase 7. These results agree with the notion that the main mode of action of statins requires use of the intrinsic pathway of apoptosis and the activation of pro-caspases [7–9]. Hence, the death pathway activated by lovastatin in HGT-1 cells is probably apoptosis.
Apoptosis characteristics in lovastatin-treated cells
Intermediates of cholesterol synthesis prevent statin-induced cell death
Because statins prevent formation of mevalonate, we investigated the putative compensatory roles of mevalonate as well as FPP, GGPP and cholesterol, the end product of the pathway, on the apoptotic response to statins. We first analysed the amount of mevalonate and cholesterol in our cell culture medium [containing 5% (v/v) foetal calf serum]. No mevalonate was detected (detection limit 1 μM), whereas 20 μM cholesterol was determined. Hence, statin treatment was performed in the absence of mevalonate, the product of HMG-CoA reductase activity. As shown in Figure 3, all compounds reduced statin-dependent apoptosis. Added mevalonate, FPP and GGPP strongly reduced apoptosis at concentrations of 50 μM, even more so than cholesterol. In addition, GGPP was the strongest inhibitor of apoptosis, with a maximum activity at a concentration of 10 μM. Of note, partial compensation of the apoptotic activity of lovastatin was observed for 10 μM added cholesterol, although the concentration of cholesterol in the culture medium was higher. This apparent discrepancy is probably due to the fact that the bioavailability of cholesterol in serum, which is essentially contained within lipoprotein particles, is much lower than that of purified cholesterol for HGT-1 cells. Taken together, these results suggest that target proteins that require prenylation (farnesylation, geranyl-geranylation), such as members of the Ras and Rho protein families, play important roles in the control of the apoptotic response in HGT-1-treated cells, as it has been demonstrated for other cell types [22,23]. It is possible that the protective effects brought about by cholesterol on the one hand, and FPP or GGPP on the other hand, depend on different mechanisms. Indeed, statins inhibit Rho cell signalling and this effect can be reversed by GGPP because GGPP provides a lipophilic anchor, which is essential for membrane attachment and biological activity of the Rho GTPases [24–26]. In contrast, cholesterol, as a major component of lipid raft caveolae, could help restore proper membrane architecture which was likely to be altered by statins [27,28].
Effect of mevalonate pathway intermediates and cholesterol on lovastatin-dependent apoptosis
Caspase levels are modulated by statins
Lovastatin induces pro-caspase levels
In order to determine the influence of statins on pro-caspases, Western blot analyses were performed after 24 h of treatment with lovastatin in several cancer cell types. As shown in Figure 4(A), all pro-caspases surveyed showed slightly higher protein levels, not only in HGT-1 cells, but also in HCT116 colon cancer (except for pro-caspase 6) and HepG2 hepatoma cells. Pro-caspase 7 induction was the most marked in all cell lines, and was highest for HGT-1 cells. It has been reported that statins could also decrease expression of the anti-apoptotic Bcl-2 and survivin proteins [7,29], an observation that we failed to make in the present study in HGT-1 cells, suggesting that an increase in pro-apoptotic proteins, rather than a decrease in anti-apoptotic proteins, was induced in response to lovastatin (results not shown). In addition, a dose–response experiment was performed in HGT-1 cells, which showed increased pro-caspase 7 levels for 5 μM lovastatin, with a maximum between 25 and 50 μM (Figure 4B).
Pro-caspase levels in three human cancer cell lines in response to lovastatin
CASP7 is a SREBP-1/2-responsive gene
To determine whether the inducibility of pro-caspases also occurred at the mRNA level, we assayed their expression by RT–PCR in response to lovastatin. However, no response was observed for caspases 3, 6 or 9, indicating that lovastatin was acting at the translational or post-translational levels for these caspases (results not shown). In contrast, caspase 7 mRNA was increased, albeit modestly (1.5–2-fold), by lovastatin following treatment with 50 μM for 16 h (Figure 5A). Analysis of caspase 7 mRNA stability by DRB (5,6-dichlorobenzimidazole riboside) treatment (80 μM) revealed that the mRNA half-life was relatively short (approx. 7 h) and was not influenced by lovastatin, indicating that lovastatin did not act to stabilize the mRNA in a post-transcriptional fashion (results not shown). Translation inhibition by cycloheximide treatment led to a 2-fold induction of caspase 7 mRNA, an effect that was not influenced by lovastatin, suggesting that the caspase 7 mRNA level was under the control of a repressor protein (results not shown), whose effect was not dependent on lovastatin. This was probably unrelated to nonsense-mediated mRNA decay since no prediction of a premature translational stop codon could be made for caspase 7 transcripts . Bioinformatics analyses showed that, among the caspase genes surveyed in the present study, only the CASP7 gene contained putative SREBP-response elements (within the proximal 1.5 kb sequence upstream of the transcriptional start site; MatInspector program at http://www.genomatix.de). In order to look for the involvement of SREBP transcription factors in the response to lovastatin, transient transfection experiments with the promoter region of the CASP7 gene were performed, which showed only a modest positive response to lovastatin or SREBPs (at most 1.5-fold) in co-transfection experiments with plasmids encoding the transcriptionally active forms of the SREBP-1 or SREBP-2 (results not shown). Although limited, this response was of the same magnitude as that observed at the mRNA level (Figure 5A). As a more relevant approach to determine the putative role of SREBP-1 or -2 on the endogenous CASP7 gene, we performed ChIP analyses focussing on the proximal promoter region, which showed the occurrence of two putative SREBP DNA-binding sites. As shown in Figure 5(B), this promoter region of the CASP7 gene was able to bind SREBPs in live cells, suggesting that these may indeed participate in the regulation of the gene. Finally, we knocked down SREBP-1 or -2 mRNAs with specific siRNAs. As shown in Figure 5(C), a strong suppression of basal caspase 7 mRNA was observed (45% and 54% decrease for SREBP-1 and SREBP-2 siRNAs respectively). As a control, FAS gene expression, a known SREBP-1-responsive gene, was reduced by 34% by SREBP-1 siRNAs (results not shown). In addition, FPPS and HMG-CoA reductase mRNAs, two recognized positive targets of SREBP-2, were also reduced by 36% and 46% with SREBP-2 siRNAs respectively (results not shown). Therefore the CASP7 gene is a true SREBP-1/2-responsive gene, and its basal expression level, i.e. in the absence of statins, is positively controlled by SREBP-1/2. CASP7 is the second caspase gene found to respond to SREBPs, as we have previously shown this to be the case for the CASP2 gene [17,18]. In addition, caspase 7 expression was decreased by approx. 30–40% with a 24 h treatment by cholesterol (78 μM), in agreement with the control exerted by SREBP overexpression on the gene (results not shown). Finally, preliminary evidence has shown that silencing S1P (site-1 protease) (a key component of the SREBP-responsive pathway) could also lead to a partial reduction of pro-caspase 7 protein levels in HGT-1 cells. The higher responsiveness of the pro-caspase 7 protein, as compared with the other pro-caspases, agrees with the notion that both transcriptional and post-transcriptional mechanisms occurred, making CASP7 a sensitive SREBP-dependent, lovastatin-response gene. Finally, suppression of SREBP-1 and SREBP-2 with siRNAs also reduced pro-caspase 7 protein levels, but did not affect pro-caspase 9 used as a control (Figure 5D). Furthermore, lovastatin responsiveness was partially maintained in the presence of SREBP-1 and -2 siRNAs, in agreement with the fact that this induction of pro-caspase 7 was afforded by both transcriptional and post-transcriptional mechanisms (Figure 5E).
Caspase 7 mRNA level in response to lovastatin, ChIP analysis and effects of SREBP-1 and -2 siRNAs in HGT-1 cells
Selection of statin-resistant cells
As an approach to derive statin-adapted cells, we grew HGT-1 cells in the continuous presence of 50 μM lovastatin (see the Materials and methods section). As stated above, no mevalonate was available to the cells as a compensatory metabolite. Hence the selection procedure was performed under conditions where the pathway was blunted. As expected, a strong apoptotic mortality was observed initially, but a few cells eventually emerged after 6 weeks and grew as novel populations (not cell clones since no limit dilution had been used), which we called L50 cells. Following culture in the absence of statin for several cell doublings, no death occurred upon re-addition of lovastatin, indicating phenotype stability. HGT-1-derived L50 cells displayed altered morphology with frequent cytoplasmic inclusions (Figure 6A). In addition, the L50 cells had a slower growth rate than HGT-1 cells (Figure 6B).
Morphology and growth of statin-sensitive and statin-resistant cells
Cell adaptation to lovastatin induces a remodelling in gene expression
To further characterize these cells, several mRNA species were analysed by RT–PCR. As shown in Figure 7(A), L50 cells had higher HMG-CoA reductase mRNA levels than HGT-1, as may be expected since HMG-CoA reductase is the primary target of statins. Strikingly, caspase 7 mRNA was overexpressed in L50 cells, whereas caspase 9 and, to a lower extent, SREBP-1 mRNAs, were reduced. The pro-caspase 7 protein level was also higher in L50 cells, whereas pro-caspase 9 was decreased (Figure 7B). In all individual HGT-1 statin-resistant cell populations, as well as in HCT116 statin-resistant colon cancer cells (Supplementary Figure S1 at http://www.BiochemJ.org/bj/420/bj4200473add.htm), the level of pro-caspase 7 was increased as compared with that of the parental, statin-sensitive cells. Hence, expression of the CASP7 gene was markedly increased in lovastatin-resistant cell populations, an effect that was associated with higher levels of pro-caspase 7. We believe that the lovastatin treatment has selected, among the parental cell populations, those cells that already expressed higher caspase 7 levels, which may have helped overcome the deprivation of the biochemical species that were lacking when the mevalonate pathway was blunted.
Gene expression levels in HGT-1 and L50 cells
ChIP assays showed that SREBPs were able to interact with the CASP7 promoter in L50 cells as they did in HGT-1 cells (see Figure 8A, which shows the same results for HGT-1 cells as Figure 5B). Caspase 7 mRNA was suppressed in L50 cells upon transfection with SREBP-1 and -2 siRNAs, as occurred for HGT-1 cells (Figure 8B). Furthermore, the level of pro-caspase 7 was also suppressed by siRNAs against SREBP-1 and/or SREBP-2 (Figure 8C). Caspase 7 mRNA half-life was increased in L50 cells (approx. 10 h) as compared with HGT-1 cells (approx. 7 h), but was not influenced by lovastatin treatment, as in HGT-1 cells (results not shown). Finally, statin responsiveness was maintained in L50 cells (Figure 8D). Hence, it appears that long-term treatment of HGT-1 cells by lovastatin has selected cells with higher levels of pro-caspase 7, which may be due to increased transcription of the CASP7 gene or, more likely, to increased stability of caspase 7 mRNA, which resulted in increased levels of protein.
SREBP control of CASP7 gene expression in statin-resistant cells
The present study identifies for the first time human CASP7 as a novel SREBP-1/2-responsive gene, similar to the CASP2 gene . The two main aspects of the present study are illustrated in Figure 9. Upon short time treatments (16–24 h), lovastatin could increase expression of caspase 7, mostly at the protein level (Figure 9A), whereas SREBPs appeared to mainly control basal expression levels of the gene. Following long-term exposure (several weeks) to lovastatin, apoptosis-resistant cells were selected, and showed stably increased levels of both caspase 7 mRNA and protein (Figure 9B). However, the apoptosis-inducing activity of lovastatin could be, at least in lovastatin-resistant cells, uncoupled from pro-caspase 7 inducibility, since inducibility was maintained in L50 cells, and was not associated with a rise in DEVDase activity, indicating that the apoptotic machinery was not constitutively active in these cells (results not shown). Overexpression of caspase 7 in L50 cells may have represented a selective advantage for these cells, separate from apoptosis, when mevalonate shortage was imposed.
Summary of the main aspects of the present study
Caspase 7 has often been considered as a surrogate to caspase 3 in apoptosis since both caspases appear to act redundantly during the last steps of the suicide programme, and caspase 7 expression could be increased in cells deficient in caspase 3 [31,32]. Yet, CASP3 knockout mice showed several abnormalities, including neuroepithelial progenitor dysfunction, but survived with some degree of perinatal lethality on the 129 genetic background. By contrast, CASP7 knockout mice were fully viable and healthy on the same background, strongly suggesting that these caspases are not fully redundant [33,34]. In addition, caspase 3 and caspase 7 have partially distinct substrate specificity, as shown by a degradomic approach  and by substrate cleavage analysis . Thus the functions of caspase 7 are unlikely to be fully interchangeable. The present results show that their response to lovastatin, and expression in L50 cells, were different, which also suggests differential regulation and, possibly, function.
It is worthy of note that L50 cells characterized in the present study grew more slowly than HGT-1 cells. How slower cell growth could be related to the overexpression of caspase 7 will require further analyses. We are currently addressing the possibility that caspase 7 may play a role in protein prenylation. Indeed, HGT-1 cells do not carry K-Ras codon 12 or 13 mutations, the most frequent Ras mutations (results not shown), indicating that the signal pathway activated by Ras is functional in these cells, and may be blunted by statins, or by adaptation to statins.
No additional caspase gene was found to respond to lovastatin, nor was predicted to be able to do so since no putative SREBP-binding sites were found by bioinformatics analysis (results not shown). Contrary to caspase 2 , we did not obtain any evidence that caspase 7 could exert a positive control on cholesterol levels in HGT-1 cells. Nevertheless, our results suggest that caspase 7, as an SREBP target, may bear a novel, unreported biological activity, beyond execution of apoptosis, which may further exemplify the fact that caspases can participate in functions unrelated to cell death .
We wish to thank Dr C. Laboisse (Université de Nantes, Faculté de Médecine de Nantes, EA Biometadys, Cedex Nantes, France) for his gift of HGT-1 cells and Dr L. Henneman and Dr H. Waterham (The Laboratory of Genetic and Metabolic Diseases, Departments of Clinical Chemistry and Paediatrics, Academic Medical Centre, Amsterdam, The Netherlands) for the mevalonate determination assays.
Dulbecco's modified Eagle's medium
fatty acid synthase
farnesyl pyrophosphate synthase
heat-shock cognate 70 stress protein
LDL (low-density lipoprotein) receptor
small interfering RNA
Laurie Gibot and Julie Follet analysed apoptosis features and the effects of compensatory metabolites, Brigitte Simon performed several RT–PCR analyses, Jean-Philippe Metges and Pierrick Auvray were involved in some aspects of the initial and final discussions, Catherine Le Jossic-Corcos performed all other experiments, except the DNA ChIP analyses (which were performed by Laurent Corcos). Catherine Le Jossic-Corcos and Laurent Corcos wrote the manuscript.
This work was supported by INSERM; the Cancéropôle Grand Ouest; the Ligue contre le Cancer (committees of Finistère, Côtes d'Armor and Cher); the FEDER (Fonds Européen de Développement Régional) funds [grant number PRESAGE 9511]; the Brittany Region [project number XCOR 5012838]; the CRITT (Centre Régional d'Innovation et de Transfert de Technologie dans le domaine de la Santé) Santé Bretagne [project number CASAC 2935]; the University of Brest (institutional university grant) and the Medical Faculty of the University Hospital of Brest (Interface contract to L. C.). L. G. was supported by a fellowship grant from Agrocampus (Rennes, France). J. F. was the recipient of a fellowship grant from the Conseil Régional de Bretagne.
C. L. J.-C. and B. S. are employees of the Ministry of Research and Education.