Lipid-modified G4-decoy oligonucleotide anchored to nanoparticles: delivery and bioactivity in pancreatic cancer cells

KRAS is mutated in >90% of pancreatic ductal adenocarcinomas. As its inactivation leads to tumour regression, mutant KRAS is considered an attractive target for anticancer drugs. In this study we report a new delivery strategy for a G4-decoy oligonucleotide that sequesters MAZ, a transcription factor essential for KRAS transcription. It is based on the use of palmitoyl-oleyl-phosphatidylcholine (POPC) liposomes functionalized with lipid-modified G4-decoy oligonucleotides and a lipid-modified cell penetrating TAT peptide. The potency of the strategy in pancreatic cancer cells is demonstrated by cell cytometry, confocal microscopy, clonogenic and qRT-PCR assays.


by NMR
. We also demonstrated that the insertion of two para-TINA (twisted intercalated nucleic acid) units in the sequence, shifts the equilibrium towards the unimolecular structure (T M = 79 °C, 100 mM KCl) 12 . The TINA-modified quadruplex called 2998 showed a strong bioactivity, as it suppressed KRAS, inhibited proliferation and activated apoptosis in Panc-1 cancer cells 12 . Furthermore, injected into a Panc-1 xenograft, 2998 inhibited tumour growth and increased the survival time of the mice by ~50% 12 . This modified oligonucleotide is expected to act through a decoy mechanism by sequestering MAZ and thus depriving the promoter of an essential transcription factor 14 . To improve the delivery of 2998, which in previous studies was given to the cells complexed to jet-PEI (PolyPlus), we designed a new strategy based on the use of palmitoyl-oleyl-phosphatidylcholine (POPC) liposomes to which we anchored noncovalently the oligonucleotide modified with a lipid moiety. Here, we demonstrate that the G4-decoy 2998 anchored to the liposomes efficiently internalize in pancreatic Panc-1 cancer cells where it reduces the metabolic activity, the clonogenicity as well as the level of KRAS transcript.

Results and Discussion
We have previously demonstrated that the G4-decoy oligonucleotide 2998 (Fig. S1) delivered with polyethylenimine (jet-PEI) activates through a decoy mechanism a strong apoptotic response in Panc-1 cells and reduces the growth of a Panc-1 xenograft in mice 12 . To improve the delivery of the G4-decoy, we have designed a transport system based on the low toxicity of palmitoyl-oleyl-phosphatidylcholine (POPC) liposomes in combination with surface attached functionalities 15 . POPC liposomes are functionalized with a cell-penetrating peptide (CPP), either the trans-activator of transcription of the human immune-deficiency virus (TAT) or the cationic octaarginine peptide (R8), and G4-decoy oligonucleotide 2998 [16][17][18][19][20] .
As the synthesis of bioconjugates between the G4-oligonucleotide and CPP is very demanding and would require a new synthesis for each new peptide or oligonucleotide used in the bioconjugate, a delivery strategy based on POPC liposomes is an attractive alternative [21][22][23] . To functionalize the liposomes a non-covalent membrane anchoring strategy for both the G4-oligonucleotide and the CPP was employed. Both functionalities, peptide and oligonucleotide, were chemically modified with a palmityl membrane anchor to allow their rapid attachment to the liposome surface [21][22][23] . The strategy is illustrated in Fig. 1. POPC liposomes are treated with  8,12 . The oligonucleotides are chemically modified as they contain 2 para-TINA insertions ( ), two locked nucleic acid modifications (A L G L ) and one insertion with two saturated palmityl chains in order to anchor the oligonucleotides to the liposomes. ODN-3 is a non-G4 oligonucleotide used as control. The sequence of the two CPP peptides (TAT and R8) is shown. They contain two lipid insertions in order to anchor them to the liposomes; (B) POPC liposomes spontaneously self-assembles into spherically closed bilayer membrane on the surface of which CPP and G4-decoy ODNs are attached through their lipid modifications ( or ); (C) Structure of para TINA (P) inserted in the G4-decoy oligonucleotides; structures of lipid insertion contained in R8 and TAT and contained in ODN1-3; (D) Proposed structure for the G4-decoys ODN-1 and ODN-2.
Scientific RepoRts | 6:38468 | DOI: 10.1038/srep38468 the lipid-modified G4-oligonucleotide and peptide that spontaneously anchor to the liposome surface 24 . As the G4-decoys are not covalently attached to the liposomes, they can move freely on the lipid surface and interact efficiently with the target proteins. The membrane anchor of the G4-decoy consists of a 3-amino-1,2-propanediol unit with two saturated palmityl chains (membrane anchor Y) 25 . We prepared three palmityl-modified oligonucleotides (Table 1). ODN-1 and ODN-2 were designed with: (i) the sequence of truncated G4-proximal comprising G-runs 2-3-4-5 12 ; (ii) two para-TINA (P) units to stabilize the unimolecular folding of the oligonucleotide 12 ; (iii) a membrane anchor Y followed by three thymidines at the 3′ end. As previously found, the additional TTT nucleotides prevent possible oligonucleotide self-aggregation and reduce the detergent properties of the lipid-modified G4-oligonucleotide [21][22][23] . Between Y and the G4-oligonucleotide a spacer of either four (A L G L TT, ODN-1) or eleven (A L G L TTATTATTA, ODN-2) nucleotides was introduced. Two nucleotides of the spacer have been replaced with locked-nucleic acids (LNA) analogues (AG→ A L G L ), to avoid possible nuclease cleavage of the G4-motif from the lipid moiety [26][27][28] . The effect of LNA modifications on the susceptibility to nuclease degradation of DNA in duplex and quadruplex conformations has been previously reported [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27]28 . The introduction of one or two terminal LNAs in the NF-kB duplex decoy was sufficient to markedly increase the oligonucleotides stability against exo-and endo-nucleases 27,28 . Similarly two LNAs at the 3′ end of 2998 showed a high stability in serum 12 . In the light of these data we designed our lipid-modified decoys with two LNAs placed outside the G4-motif to avoid a possible effect of the sugar modification on the quadruplex structures 29,30 . As for ODN-3, it bears the same modifications as ODN-1 and ODN-2 but was designed with a random sequence that does not allow any folding (this oligonucleotide was used as a control). A similar strategy was used for the lipid modification of TAT-or R8-derived peptide. The two palmityl-modifications (X) were incorporated close to the C-terminus, followed by three additional glutamic acid residues to avoid self-aggregation of the peptide. At the same time, the three E residues provide additional negative charges and reduce the peptide's net positive charge from 7+ to 4+ thus decreasing the toxicity usually associated with strongly positive charged peptides (Fig. 1). The molecular weight of the synthesized lipid-modified oligonucleotides and peptides, after purification, was checked by MALDI-TOF mass spectrometry. The difference between theoretical and experimental masses was in general < 0.01% ( Table 1).
Emerging studies indicate that the more homogeneous the nanoparticles are, the better their performance 31,32 . So, to obtain liposomes with a homogeneous diameter, the solution was extruded 10 times through a double polycarbonate filter with pore size of 50 nm using compressed N 2 (20-40 bar) [21][22][23] . The sizes of the liposomes were estimated by nanoparticle tracking analysis (NTA). The liposomes for imaging experiments have been marked with cyanine dye (Cy5), which was encapsulated in the liposomes (Fig. S2). The plots of Fig. 2B show that POPC liposomes have slightly different sizes depending whether they contain or not Cy5: diameter = 70 nm with Cy5; diameter = 85 nm without Cy5. The average amount of incorporated dye was ~0.5 mol % Cy5 per liposome.
Next, we interrogated if the G4-decoy 2998 maintains its folded structure also when it is anchored to the liposome surface through its two palmityl chains inserted close to the oligonucleotide 3′ end. To address this question we performed CD experiments which showed that ODN-3, free or attached to the liposomes, gave the same spectrum with a maximum at 278 nm and a minimum at 250 nm, which is indicative of an unstructured oligonucleotide, in keeping with the fact that it lacks a G4-motif (Fig. 3). By contrast, ODN-1 and ODN-2, that fold into a G4-DNA structure, exhibit an enhanced ellipticity at 260 nm typical of a parallel or mixed paerallel/ antiparallel G-quadruplex 33 . Note that 2998 anchored to the liposome surface shows a 260 nm-ellipticity slightly stronger than that exhibited by the liposome free G4-decoy. This suggests that the oligonucleotide assumes its typical folded structure more efficiently when the alkyl chains are bound to and thus sequestered by the lipid bilayer of the liposomes.
The bioactivity of the designed liposomes has been analysed mainly in Panc-1 cells, in which KRAS carries one of the most common mutation detected in patients affected by pancreatic adenocarcinoma (G12D, 12Gly→ Asp) 34 . In order to evaluate their cell-penetrating capacity, POPC liposomes were marked with cyanine 5 (Cy5), which emits an intense fluorescence peak at 660 nm upon excitation at 540 or 620 nm. Cy5 was encapsulated in the  Table 1. Modified G4-decoy oligonucleotides and CPP-modified peptides (TAT or R8). P = para-TINA insertion; Y = DNA lipid modification ; X = peptide lipid modification.
liposomes by mixing the dye with POPC followed by extrusion and removal of excess dye by dialysis. Figure 4A shows a typical FACS analysis of Panc-1 cells treated for 1 and 4 h with Cy5-marked liposomes functionalized with ODN-1 (96·ODN-1 per liposome) (L) and CPP [752·(TAT or R8) per liposome] (L/TAT or L/R8). It can be seen that without CPP, the liposomes with ODN-1 were poorly taken up by the cells, as indicated by a modest shift to the right of the fluorescence peak (from FL-3~ 3 to FL- 3~ 14). In contrast, when the liposomes were functionalized with both ODN-1 and CPP, their capacity to enter into the cells significantly increased. After 1 h of incubation, TAT-liposomes showed a fluorescence peak at FL-3~ 200, while at 4 h the peak occurred at FL-3~ 170, indicating that after an initial uptake the intracellular concentration of TAT-liposomes slightly decreased (probably due to some dissociation of the encapsulated Cy5). R8-liposomes entered into Panc-1 cells in a more complex way. While the cells treated with TAT-liposomes form a uniform population typified by a Gaussian peak of fluorescence, suggesting the presence of only one mechanism of transport, R8-liposomes instead give rise to a more heterogeneous cell population, as the fluorescence of the treated cells showed a broader peak after 1 and 4 h of incubation, which reflects more than one entry pathways: may be receptor-mediated endocytosis and  direct fusion with cell membrane. Panel B shows the liposome uptake in pancreatic BxPC-3 cancer cells, which contrarily to Panc-1 are not KRAS mutated. The uptake in these cells is less efficient and does not seem to increase with CPP, suggesting a significant heterogeneity of pancreatic cancer cells. FACS analysis was performed also with MIA PaCa-2 cells, which bear a point mutation in exon 1, G12C (12Gly→ Cys). The results are reported in Fig. S3, showing that POPC liposomes enter efficiently in this cancer cell line. Together, our data indicate that the liposomes penetrate more efficiently into KRAS-mutated cell lines. Next, we examined the liposome uptake by confocal microscopy (Fig. 5). Panels A-C show Panc-1 viable cells stained with SYTO-16, a vital dye increasing its green fluorescence quantum yield upon binding to DNA and RNA 35 . SYTO-16 incubated for 2 h with the cells stains the nucleus in green much more than the cytoplasm, whose fluorescence appears slightly punctuated due to the binding of the dye to mitochondrial DNA. When Panc-1 cells were treated for 24 h with liposomes functionalized only with TAT and ODN-1 and for the last 2 h of incubation with SYTO-16 as well, a strongly punctuated fluorescence was visible in the cytoplasm, mainly in the perinuclear region (Panels D-F). This is due to the binding of SYTO-16 to the DNA anchored to the liposomes. Note that the nucleus appears now less green than in panel A because most SYTO-16 binds to the G4-decoy ODN-1 anchored on the liposomes surface (142·ODN-1 per liposome). Expectedly, the red channel did not give any signals as the liposomes were not marked with Cy5. Panels G-I show Panc-1 cells treated for 24 h with liposomes functionalized with TAT, Cy5 and with SYTO-16 for 2 h. As in panel A, the nuclei look more stained in green than the cytoplasm, where in this case the punctuated fluorescence is more visible. As expected, the red channel shows the distribution of the liposomes marked with Cy5. The liposomes are localized in the cytoplasm. As the red spots do not co-localize with the faint green spots, we can conclude that the liposomes do not internalize in the mitochondria. In panels L-N, we show Panc-1 cells treated with liposomes functionalized with TAT, Cy5 and ODN-1 (96·ODN-1 per liposome) for 24 h and with SYTO-16 for the last 2 h of incubation. As observed in panel D, most SYTO-16 is bound to ODN-1 anchored to the liposomes, so the nucleus appears poorly stained, while the cytoplasm where the liposomes accumulate appears strongly stained in green. As the liposomes are now also marked with Cy5, the red channel shows a fluorescence that co-localize with the green fluorescence of the cytoplasm. A close inspection of the micrographs shows that a small fraction of liposomes functionalized with ODN-1 is also localized in the nucleus. The confocal analyses performed with the various formulations of liposomes indicate that these nanoparticles accumulate primarily in the cytoplasm of pancreatic cancer cells. The above confocal experiments were carried out with viable cells. We also analysed the uptake in viable BxPC-3 pancreatic cancer cells (Fig. 6). Panels A, B and C show the cells treated with liposomes loaded with TAT and Cy5 for 24 h and with SYTO-16 for the last 2 h of the incubation. As observed with Panc-1 cells, the liposomes stained in red appear clearly localized in the cytoplasm (only a small fraction of liposomes is marked with Cy5). Panels D, E and F show the cells treated with liposomes functionalized with TAT, Cy5 and ODN-1 (96 ODN-1 per liposome). As with Panc-1 cells, SYTO-16 binds to ODN-1 bound to the liposomes and thus the cytoplasm appears strongly punctuated in green. The confocal images of viable MIAPaCa-2 are shown in Fig. S4. In addition, we have also analysed the uptake of Panc-1 cells fixed on glass, we obtained results in keeping with the analysis conducted with the viable cells (Fig. S5).
As G4-decoy 2998 binds to proteins recognizing G4-proximal (PARP-1, Ku70, hnRNP A1, MAZ) 9 , it inhibits KRAS transcription by a decoy mechanism. As a consequence, the proliferation and clonogenic potential of Panc-1 cells are reduced by the G4-decoy oligonucleotide 12 . To investigate the clonogenic effects of lipid-modified 2998 anchored to POPC liposomes, we first treated Panc-1 cells (bearing the point mutation G12D in exon 1 34 ) with the liposomes functionalized with TAT and ODN-1 (142 ODN-1/liposome). We started the experiment with one single treatment. After 13 days of incubation, we observed that the G4-decoy did not produce the expected inhibition of colony formation (not shown). We therefore performed two treatments, at day 2 and 5, as previous experiments showed that in this way 2998 exhibited activity in Panc-1 cells 12 . The experiment outlined is shown in Fig. 7A. The cells were treated twice with TAT-or R8-liposomes, loaded with ODN-1, ODN-2 or ODN-3. 13 days after the second treatment, the colonies formed were stained and counted. The results reported in the histogram showed that TAT-liposomes loaded with G4-decoy ODN-1 or ODN-2 strongly reduced colony formation to ~10% of the control (Panc-1 cells untreated or treated with ODN-3) (Fig. 7B). R8-liposomes reduced colony formation to only 60% of the control: a finding that may reflect the complexity of their uptake suggested by the FACS data (Fig. S6). To know if the observed strong inhibition of colony formation mediated by the G4 decoys is cell line dependent, we extended our analysis to other two adenocarcinoma pancreatic cancer cells: MIA PaCa-2 carrying the G12C (12Gly→ Cys) mutation in exon 1/codon 12 and BxPC-3 carrying wild type codon 12/exon 1 34 . The results reported in Fig. 8 show that the G4-decoys ODN-1 and ODN-2, but not the control ODN-3, significantly decrease the clonogenic potentials also in these cells, both in terms of number and size of colonies. This demonstrates that the G4-decoys anchored to the liposomes did not lose the anticlonogenic activity that was observed with the free G4-decoys delivered with jet-PEI 11 . As a next step, we determined, by quantitative RT-PCR, the effect of TAT-liposomes loaded with ODN-1, ODN-2 and ODN-3 on KRAS mRNA (Fig. 9A). We observed that the G4-decoy oligonucleotides ODN-1 and ODN-2 reduced the level of mRNA to about 50% of the control (mRNA level in untreated or ODN-3 treated cells), suggesting that the different length of the spacers between the G4 sequence and the lipid chains does not affect the decoy activity. Oligonucleotide ODN-3, which is unable to fold into a G-quadruplex, showed no inhibitory effects on KRAS transcription. The results are in agreement with a previous study carried out with 2998 delivered to the cells with PEI 12 . As KRAS is an oncogene that reprograms the metabolism of cancer cells 6,7 , we examined the metabolic activity in Panc-1 cells untreated or treated with the liposomes. The metabolic activity of Panc-1 cells was measured with resazurin: viable cells with an active metabolism reduce resazurin into resorufin, which is a pink and fluorescent molecule 36 . It can be seen that ODN-1 and ODN-2 reduce the metabolic activity to ~50% of the control (untreated or ODN-3-treated cells) (Fig. 9B). As we previously demonstrated, the G4-decoy oligonucleotide 2998 reduces both cell growth and clonogenic potential by activating apoptosis. We therefore investigated if this mechanism is also activated when 2998 is delivered with TAT-liposomes. We measured the activation of caspases 3/7 and found indeed that ODN-1 and ODN-2, but not ODN-3, activate the caspases in Panc-1 cells (Fig. 9C). This was also observed by confocal microscopy analyzing living cells treated with TAT-liposomes loaded with Cy5/ODN-1. It can be seen that some cells are completely red due to a bulky internalization of liposomes. Note that some completely red cells show the typical signs of apoptosis (indicated by arrows): bubbling and shrinkage into cell fragments (Fig. 9D).
In conclusion, in this article we have demonstrated that a G4-decoy oligonucleotide with anti-KRAS activity can be delivered with POPC liposomes marked with TAT or R8 peptide. Both functionalities, G4-decoy and TAT peptide, bear two palmityl chains that allow their anchoring to the liposome surface. As the lipid-modified G4-decoys anchored to the liposomes display their bioactivity in pancreatic cancer cells, POPC liposomes may be suitable carriers for in vivo delivery of therapeutic oligonucleotides including the G4-decoys as well as antisense and miRNAs. Work is in progress in our laboratories along this direction.

Experimental section
Synthesis of modified oligonucleotides. Oligonucleotides were synthesized on an ExpediteTM 8900 nucleic acid synthesis system (Perceptive Biosystems Inc.). The syntheses were performed on a 1.0 μ mol scale on GE Healthcare Custom Primer Support TM T40s using standard conditions for automated synthesis with DCI as activator. However, the lipid modified phosphor amidite was dissolved in 2:1 DCE:MeCN at a concentration of 0.1 M, 42 ® was used as activator (Proligo reagent/Sigma-Aldrich) and the coupling time was increased to 20 min.
The coupling time for TINA was 5.2 min. The DMT protecting group on the last nucleotide in the sequence was removed. After deprotection and cleavage from the solid support using standard conditions (conc. NH 3 (aq.) over night at 55 °C), the oligonucleotides were purified by HPLC. Purification was performed using a DIONEX Ultimate 3000 with a DIONEX Acclaim ® C18 3 μ m 300 Å reversed phase column with UV detection at 260 nm. A flow rate of 1 mL/min. and a column temperature of 50 °C using the following gradient program: 2 min. isocratic with 0.05 M triethyl ammonium acetate (TEAA) followed by a 8 min. linear gradient to 40% 1:3 H 2 O:MeCN and a 20 min. linear gradient to 100% 1:3 H 2 O:MeCN which was continued isocratic for 10 min., was used. After HPLC purification, the oligonucleotides were desalted using a NAP 10 column (Sephadex G25 grade). Mass spectra of oligonucleotides were recorded using a Bruker Daltonics microflex LT MALDI-TOF.  Preparation of POPC liposomes. Liposomes were extruded using a LIPIX Extruder, Northern Lipids.
Nanoparticle Tracking Analysis. The sizes of the liposomes were determined using Nanoparticle Tracking Analysis (NTA). NTA measurements were performed with a NanoSight LM10-HS equipped with an Andor Lucas EMCCD camera, a LM14 temperature controller and a laser diode operated at 404 nm. The data was analysed using the NanoSight NTA 2.1 software.
Due to the relatively low particle concentration necessary for NTA (10 8 -10 9 particles/mL), the samples with dye-labelled liposomes were diluted 250000-fold and the unlabelled liposomes 10000-to 100000-fold with phosphate buffer (10 mM NaH 2 PO 4 . 2 H 2 O, 5 mM Na 2 HPO 4 , 140 mM Na + , pH 7.4). All measurements were carried out in duplicate at 23 °C. Circular dichroism spectra. CD spectra were obtained using the same samples which was used for T m -measurements and were collected at 37 °C using a JASCO J-815 CD Spectrometer and cuvette with a 2.00 mm path length. The spectra have been collected in phosphate buffer and the oligonucleotide concentration was 2.0 μ M for both free oligonucleotides and oligonucleotides anchored to POPC liposomes (142 ODN/liposome). The measurements were recorded from 200 to 350 nm at a scanning speed of 50 nm/min, a data pitch of 0.2 nm and a response time of 4 s. The final spectra are the average of three measurements. For confocal images, 4.5 μl of liposome mix were added to 1 μl cell medium and the final solution used to treat the cells. For the other experiments (FACS, clonogenic assays, apoptopsis, RT-PCR and metabolic activity) the liposome mix was obtained by adding 0.36 nmol oligonucleotide in 100 μl buffer (POPC and peptide were scaled down accordingly). Then 100 μl of liposome mix were in this case added to 0.5 μl cell medium and the final solution used to treat the cells. After 48 h the treated cells of each well were divided in 3 parts and seeded in 3 wells. After 24 h a second transfection following the same protocol was performed. This protocol was used for the qRT-PCT, metabolic activity, clonogenic and apoptosis assays. Real-time PCR multiplex reactions were performed with 1 x Kapa Probe fast qPCR kit (KAPA Biosystems, Wilmington, MA, USA) for KRAS and housekeeping genes β 2-microglobulin and HPRT, 1.0 μ l of cDNA in 10 μ l final and primers/probes at the following concentrations: for KRAS, the probe was FAM-TACTCCTCTTGACCTGCTGTG-BHQ1 (accession No. NM_033360, from 352 to 372, 90 nM), the sense primer was 5′-C GAATATGATCCAACAATAGAG (from 271 to 292, 180 nM) and the antisense primer was 5′-ATGTACT GGTCCCTCATT (from 379 to 396, 180 nM). For β 2-microglobulin accession n. NM_004048 probe ROX-TATGCCTGCCGTGTGAACC-BHQ2 (from 352 to 370, 60 nM), the primer sense was 5′-CCCCACTGAAAAAGATGA (from 333 to 350, 100 nM), the primer antisense was 5′-CCATGATGCTGCTTACAT (from 415 to 432, 100 nM). For HPRT accession n. NM_000194 probe 5′-Cy5-CTTGCGACCTTGACCATCTT-BHQ2 (from 633 to 652, 180 nM), the primer sense was 5′-CTTGATTGTGGAAGATATAATTG (from 557 to 575, 210 nM), the primer antisense was 5′ -TATATCCAACACTTCGTGG (from 672 to 690, 230 nM). The PCR cycle was: 3 min at 95 °C, 50 cycles 10 s at 95 °C, 60s at 58 °C. PCR reactions were carried out with a CFX-96 real-time PCR apparatus controlled by an Optical System software (version 3.1) (BioRad Laboratories, CA, USA). KRAS mRNA was normalized with the two housekeeping genes.

RNA Extraction and
Metabolic activity and clonogenic assays. The metabolic activity (MA) of Panc-1 cells was measured on a 96-well plate where each well, containing 10 4 cells, was transfected twice with the POPC liposomes loaded with ODN (142 ODN/liposome). The MA was measured 72 h after the second transfection by a resazurin assay: 25 μ M resazurin was added to the cell medium, and the fluorescence was measured after 1 h (Ex 535 nm; Em 590 nm) with a spectrofluorometer EnSpire 2300 Multilabel Reader (Perkin Elmer).
For colony-forming assay, Panc-1, BxPC-3 and MIAPaCa-2 cells were transfected with POPC liposomes loaded with ODN (142 ODN/liposome). 18 h following the second transfection, the wells were treated with trypsin and one third of the volume of each well was seeded on 60-mm diameter plate and the cells were let to grow under normal culture conditions. After 13 days, the cells were stained with 2.5% methylene blue in Scientific RepoRts | 6:38468 | DOI: 10.1038/srep38468 50% ethanol for 10 min. Colonies of > 50 cells were counted with Image Quant TL software (Image Scanner, Amersham). Apoptosis assays. Caspase assay was performed with Apo-ONE TM Homogeneous Caspase-3/7 Assay For confocal experiments with glass-fixed Panc-cells, we plated (3 × 10 5 ) on a coverslip placed in a Petri dish (diameter 35 mm) and after 24 h treated with liposomes (36 nmol POPC) loaded with 0.8 nmol ODN-1 (380 ODN-1 per liposome), TAT (0.7 nmol) and Cy5 (0.7 nmol) for 4 h. The cells were treated also with SYTO 14 for 1 h. The cells have been washed twice with PBS and fixed with 3% paraformaldehyde (PFA) in PBS for 20 min. After washing with 0.1 M glycine, containing 0.02% sodium azide in PBS to remove PFA and Triton X-100 (0.1% in PBS), the nuclei have been stained for 5 min with Hoechst (3 ng/μ L in PBS). The cells were analysed on a Leica TCS SP1 confocal imaging system.