Adapting AlphaLISA high throughput screen to discover a novel small-molecule inhibitor targeting protein arginine methyltransferase 5 in pancreatic and colorectal cancers

Abstract

Pancreatic ductal adenocarcinoma (PDAC) and colorectal cancer (CRC) are notoriously challenging for treatment. Hyperactive nuclear factor κB (NF-κB) is a common culprit in both cancers. Previously, we discovered that protein arginine methyltransferase 5 (PRMT5) methylated and activated NF-κB. Here, we show that PRMT5 is highly expressed in PDAC and CRC. Overexpression of PRMT5 promoted cancer progression, while shRNA knockdown showed an opposite effect. Using an innovative AlphaLISA high throughput screen, we discovered a lead compound, PR5-LL-CM01, which exhibited robust tumor inhibition effects in both cancers. An in silico structure prediction suggested that PR5-LL-CM01 inhibits PRMT5 by binding with its active pocket. Importantly, PR5-LL-CM01 showed higher anti-tumor efficacy than the commercial PRMT5 inhibitor, EPZ015666, in both PDAC and CRC. This study clearly highlights the significant potential of PRMT5 as a therapeutic target in PDAC and CRC, and establishes PR5-LL-CM01 as a promising basis for new drug development in the future.

Keywords: AlphaLISA; PRMT5; colorectal cancer; pancreatic ductal adenocarcinoma; small-molecule inhibitor.

Figure 1. PRMT5 is overexpressed in pancreatic and colorectal cancer

A. and B. (Left panels), Western blots, showing that A. PRMT5 expression is higher in PDAC cells (AsPC1, MiaPaCa2 and PANC1) as compared to control pancreatic cells HPNE. B. PRMT5 expression is higher in CRC cells (HT29, HCT116 and DLD1) as compared to control colon cells FHC. β-actin was used as a loading control. A. and B. (Right panels), IHC staining, showing that A. PRMT5 expression was higher in PDAC tumor tissues as compared to the normal tissue. B. PRMT5 protein expression was higher in the inflammation, polyp, advanced stages of CRC and the metastatic stage as compared to the normal colon tissue.

Figure 2. PRMT5 is important for cell proliferation, anchorage-independent growth and migration of PDAC and CRC

A. Western blot, confirming stable PRMT5 overexpression and shPRMT5 knockdown in PANC1 and HT29 cells. B. Cell proliferation assay, showing that cell proliferation was significantly higher in the WT-PRMT5 cell lines, while shPRMT5 cells exhibited quite opposite effect in both PANC1 and HT29 cell lines. C. Anchorage-independent growth assay, showing that anchorage-independent growth was significantly higher in the WT-PRMT5 cell lines, while dramatically reduced in the shPRMT5 cells. Both colony size and number are quantified at the bottom of the corresponding panels. The data represent the means ± standard deviation (S.D.) for three independent experiments. *P < 0.05 vs. control (Ctrl) group. D. Cell migration assay, showing that cell migration was significantly higher in the WT-PRMT5 overexpression cells, while significantly reduced in the shPRMT5 cells. Upper panels, representative pictures in 20X magnification. Lower panel, quantification for the change in migration. The data represent the means ± S.D. for three independent experiments. *P < 0.05 vs. Ctrl.

Figure 3. Discovery of PR5-LL-CM01 as a novel PRMT5 inhibitor through PRMT5-specific AlphaLISA HTS technique

A. Schematic illustration of the AlphaLISA technique. Biotinylated histone H4 is incubated with PRMT5 and methyl donor SAM. PRMT5 symmetrically dimethylates H4 on its arginine 3, a product that is recognized by Acceptor beads. Following the binding of Donor beads, the chemiluminescent signal is emitted and detected by the EnVision^® Reader. The intensity of the Alpha signal is proportional to the PRMT5 activity. B. (Upper panel) Western blot, showing Flag-tagged WT-PRMT5 was overexpressed in the 293-WT-PRMT5-Flag cells as compared to the control 293 cells. 293-WT-PRMT5-Flag cells were used to purify PRMT5 for AlphaLISA assay. (Lower panel) Z’-plot of AlphaLISA assay, showing that Alpha signal with PRMT5 (red dots) is significantly higher than those without PRMT5 (blue dots). The mean values are represented with solid horizontal lines in the center of each sample group. Z’ value for our experiment was 0.6. C. Structure of PR5-LL-CM01, highlighting the core structure of a pyrozolo-pyrimidine group (red box), and three surrounding groups A, B and C, respectively. D. Calculation of IC₅₀ of PR5-LL-CM01 using AlphaLISA, showing PR5-LL-CM01 concentration-dependent decrease of Alpha signal. The IC₅₀ was ~7.5 μM. E. Co-IP-Western blot, showing that treatment with 20μM PR5-LL-CM01 inhibited p65 methylation, a PRMT5 substrate, in 293-WT-p65-Flag cells. Flag beads were used to pull down WT-p65-Flag and samples were probed with anti-symmetric dimethyl arginine motif Ab (sdme-RG Ab).

Figure 4. PR5-LL-CM01 is a potent inhibitor of PRMT5 in PDAC and CRC cells

A. MTT assay in PDAC cells (PANC1, MiaPaCa2 and AsPC1), showing that cell viability decreased significantly in presence of increasing concentrations of PR5-LL-CM01. B. MTT assay in CRC cells (HT29, HCT116, and DLD1), showing dramatic decrease of cell viability in presence of increased concentrations of PR5-LL-CM01. C. Table, summarizing the IC₅₀ values for PR5-LL-CM01 in PDAC and CRC cells, respectively.

Figure 5. Treatment with PR5-LL-CM01 significantly inhibited NF-κB activation in PDAC and CRC cells

A. NF-κB luciferase assay, showing that overexpression of WT-PRMT5 led to NF-κB activation, while shPRMT5 resulted in quite opposite effect in both PANC1 and HT29 cells. *P < 0.05 vs. Ctrl. B. Luciferase assay, showing a decrease in NF-κB activation with increasing concentrations of PR5-LL-CM01 in PANC1 (left panel) and HT29 cells (right panel). A much higher concentration of EPZ015666 is needed in order to reach similar level of NF-κB inhibition as that of PR5-LL-CM01. The data represent the means ± S.D. for three independent experiments. *P < 0.05 vs. Untreated group. C. qPCR analysis, showing that overexpression of PRMT5 significantly enhanced IL-1β-triggered NF-κB target genes (TNFα and IL8) expression, while shPRMT5 exhibited quite opposite effect, in both PANC1 and HT29 cells. *P < 0.05 vs. Ctrl; ^#P < 0.05 vs. Ctrl+IL-1β-treated group. D. qPCR analysis, showing that treatment with PR5-LL-CM01 dramatically decreased TNFα and IL8 expression, in both PANC1 and HT29 cells. The data represent the means ± S.D. for three independent experiments. *P < 0.05 vs. Untreated group.

Figure 6. PR5-LL-CM01 displayed significant anti-tumor effect in vivo

A. No significant changes in body weight were observed over the course of treatment in either PANC1 or HT29 model after treatment with 20mg/kg of PR5-LL-CM01. (*P < 0.05, n = 4). B. Tumor efficacy study for PANC1 or HT29 cells which were subcutaneously implanted in NSG mice. Inhibition of tumor growth was observed upon treatment with 20mg/kg of PR5-LL-CM01 intraperitoneally for 3X/week, as compared to the vehicle control. (*P < 0.05, n = 4).

Figure 7. In silico prediction of the binding of PR5-LL-CM01 and EPZ015666 to PRMT5

A. left panel, In SAM (purple color)-bound condition, the binding sites for PR5-LL-CM01 (turquoise color) and EPZ015666 (orange color) on PRMT5 overlap to a great extent, suggesting that these two compounds inhibit PRMT5 through largely similar group of binding sites. Right panel, in Apo-PRMT5 condition, the binding of EPZ015666 remains the same as that in the left panel, however, PR5-LL-CM01 has shifted dramatically from its left panel’s position (red dotted line) to the new position (pink dotted line), which is same as SAM's position in the left panel. B. Table, listing the binding affinities of PR5-LL-CM01-PRMT5 interactions in SAM bound and Apo-PRMT5 conditions. PR5-LL-CM01 showed favorable binding energies under both conditions, with Apo-PRMT5 condition having a slightly stronger binding affinity (-7.911 kcal/mol). C. Ligand affinity map in the SAM-bound condition, depicting the PRMT5 residues interacted with PR5-LL-CM01. Different bonds or charges are symbolized on the right side of the figure. Key site, Glu444, is circled in red dotted line. D. Binding residue distribution diagram, illustrating the PRMT5 residues that potentially bind to PR5-LL-CM01 alone (red font), EPZ015666 alone (blue font), or both (green font). E. Hypothetical model, proposing that PRMT5 overexpression promotes cancer phenotype, whereas inhibition using PR5-LL-CM01 impedes PDAC and CRC progression by inhibiting NF-κB activation.