MAZ-binding G4-decoy with locked nucleic acid and twisted intercalating nucleic acid modifications suppresses KRAS in pancreatic cancer cells and delays tumor growth in mice

Abstract

KRAS mutations are primary genetic lesions leading to pancreatic cancer. The promoter of human KRAS contains a nuclease-hypersensitive element (NHE) that can fold in G4-DNA structures binding to nuclear proteins, including MAZ (myc-associated zinc-finger). Here, we report that MAZ activates KRAS transcription. To knockdown oncogenic KRAS in pancreatic cancer cells, we designed oligonucleotides that mimic one of the G-quadruplexes formed by NHE (G4-decoys). To increase their nuclease resistance, two locked nucleic acid (LNA) modifications were introduced at the 3'-end, whereas to enhance the folding and stability, two polycyclic aromatic hydrocarbon units (TINA or AMANY) were inserted internally, to cap the quadruplex. The most active G4-decoy (2998), which had two para-TINAs, strongly suppressed KRAS expression in Panc-1 cells. It also repressed their metabolic activity (IC50 = 520 nM), and it inhibited cell growth and colony formation by activating apoptosis. We finally injected 2998 and control oligonucleotides 5153, 5154 (2 nmol/mouse) intratumorally in SCID mice bearing a Panc-1 xenograft. After three treatments, 2998 reduced tumor xenograft growth by 64% compared with control and increased the Kaplan-Meier median survival time by 70%. Together, our data show that MAZ-specific G4-decoys mimicking a KRAS quadruplex are promising for pancreatic cancer therapy.

Figure 1.

(A) Sequence of the NHE present in the human KRAS promoter upstream of the transcription start site. NHE contains six guanine repeats and three quadruplex-forming motifs, named 32R (repeats 1–6), 32R-3n (repeats 2–6) and 21R (repeats 1–4); (B) CD of the quadruplex-forming sequences at 20°C and 90°C; (C) sequence homology between human and murine NHEs. Human NHE contains two binding sites [(GGG(A/C)GG] for the transcription factor MAZ at the 5′- and 3′-ends.

Figure 2.

(A) Mobility shift assay showing the binding of recombinant MAZ to KRAS quadruplexes of NHE. Increasing amounts of MAZ (1, 2 and 4 μg) were incubated for 30 min in binding buffer with 10 nM radiolabeled G-quadruplex and run in 5% PAGE in TB buffer (500 V, 3 h, 20°C). The white arrows indicate the MAZ-G4 complexes. The radiolabeled quadruplexes have also been incubated with nuclear extract from Panc-1 and HeLa cells (3 μg). They form two main G4-protein complexes (P:G4). Bovine serum albumin (4 μg) was used as a control. (B) Left panel shows the level of MAZ mRNA in Panc-1 cells transfected with pCMV-MAZ; right panel shows KRAS mRNA level in the same cells. C = cells treated with an empty plasmid, T = treated cells. (C) Left panel shows the level of MAZ mRNA in Panc-1 cells in which MAZ was silenced with siRNA; right panel shows KRAS mRNA level in the same cells. C = cells treated with control siRNA, T = treated cells. Student’s t-test, **P < 0.01. (D) Filter-binding assay showing the binding of recombinant MAZ to quadruplex 32R-3n, in 50 and 100 mM KCl. Top membrane (nitrocellulose) shows MAZ [or control proteins, trypsinogen (T), ovalbumin (O)] bound to the quadruplex; bottom membrane (nylon+) shows unbound radiolabeled quadruplex. The fraction of quadruplex 32R-3n bound to MAZ is plotted against the MAZ concentration. The binding curve relative to MAZ was best-fitted to a standard binding equation. A K_D of 320 ± 2 nM was obtained.

Figure 3.

Chemical insertions introduced in the quadruplex-forming sequence 32R-3n. The primary structures of the designed G4-decoys are reported in Table 1.

Figure 4.

(A) Twenty per cent PAGE in TB, pH 8, 50 mM KCl, 20°C (200 V, 8 h) of the designed G4-decoys (5 μM), including the control sequences 5153 and 5154. Before loading, the oligonucleotides have been incubated overnight in the buffer containing 100 mM KCl. The 32R-3n was loaded after 1 h or overnight incubation. (B) CD spectra in 50 mM Tris–HCl, pH 7.4, 100 mM KCl, 2-mm cuvette, of the designed G4-decoys at 20°C. P = para-TINA; O = ortho-TINA, Y = AMANY, M = MADS. R is the CD ratio between the signals at 290 over 265 nm. (C) Percentage of MA compared with control (NT) of Panc-1 cells treated with 600 nM G4-decoys and control oligonucleotides, carried out 48 h after oligonucleotide treatment. NT = non-treated cells. (D) EMSA showing the capacity of the designed decoys to compete the binding of MAZ to quadruplex 32R-3n. All lanes contained 8 nM radiolabeled 32R-3n and 4 μg of recombinant MAZ. Lanes 1, 10, 11 and 24 show the DNA–protein complex in the absence of a decoy competitor. The other lanes show that the DNA–protein complex is competed by the decoys, 10- and 50-fold in excess to radiolabeled 32R-3n. Analysis: 5% PAGE in TB, 500 V, 3 h, 20°C. Before being used, the competitor decoys and probe 32R-3n were annealed in 100 mM KCl.

Figure 5.

(A) The 32R-3n sequence showing the bases methylated by DMS (black square, methylated; white square, protected from methylation) and photocleaved by TMPyP4/light (red square). Sequence variants of 32R-3n and decoy 2998 are also shown. (B) Eighteen per cent PAGE in the presence of 50 mM KCl (200 V, 8 h, 20°C); CD of 5 μM oligonucleotides in 100 mM KCl. (C) Putative mixed p/ap structure of 2998; CD-melting curves of 2998 and decoy variants. All experiments performed in 50 mM Tris–HCl, pH 7.4 and 100 mM KCl.

Figure 6.

(A) Level of KRAS mRNA in Panc-1 cells treated with 600 nM G4-decoy and control oligonucleotides, measured 24 h after decoy transfection in the presence of jetPEI. KRAS mRNA was measured with respect to housekeeping genes β2-microglobulin and HPRT. Ordinate reports the percentage of KRAS transcript, i.e T/C × 100 where T is (KRAS transcript)/(β2-microglobulin and HPRT transcripts) in the decoy-treated cells, whereas C is (KRAS transcript)/(β2-microglobulin and HPRT transcripts) in the decoy-untreated cells. (B) Western blot showing the level of p21^RAS at 48 and 72 h in Panc-1 cells after transfection with decoy and control oligonucleotides, in the presence of jetPEI. The data have been normalized to β-actin. The band intensities have been measured with ChemiDOC XRS apparatus (BioRad). Right panel shows a histogram reporting the average values of three independent western blots (12% SDS–PAGE, 1 h, 180 V).

Figure 7.

Colony-forming assay in Panc-1 cells treated with decoy and control oligonucleotides (32R-3n, 2998, 3044, 5153 and 5154). Colonies have been fixed and stained with methylene blue, 7 days after cell transfection. The histogram shows the percentage of colony formation of the treated cells compared with untreated cells. Difference from control is supported by Student’s t test, **P < 0.01.

Figure 8.

(A) Apo ONE assay showing caspase 3/7 activation in Panc-1 cells 30 h after treatment with decoy and control oligonucleotides. (B) Cell cycle analysis of Panc-1 cells 24 h after treatment with decoy and control oligonucleotides. The cells were stained with propidium iodide and analyzed by flow cytometry. (C) Propidium iodide/annexin V assay for the detection of early and late apoptotic Panc-1 cells, 40 h after transfection with the control and decoy oligonucleotides.

Figure 9.

(A) Effect of decoy and control oligonucleotides intratumorally injected in SCID mice bearing a subcutaneous Panc-1 xenograft. The ordinate axis reports the growth of the tumor xenograft (mg). The oligonucleotide treatment (2 nmol/mouse) was performed three times at Days 1, 6 and 11 (t1, t2 and t3). From Day 13, the Panc-1 xenograft in the mice group treated with 2998 grew more slowly than in the non-treated mice group or mice groups treated with controls 5153 and 5154 (P < 0.001). (B) Kaplan–Meier curves showing the effect of decoy 2998 compared with the untreated group or the groups treated with 5153 or 5154. The median survival time of the group treated with 2998 is 103.5 days, statistically higher than the median survival time of the control groups (71 and 71.5 days) (P < 0.006).